As its a timelapse i cropped out the ui as it would be pretty useless to look at.
Not the best model nor am i an amazing modeller but it should show that the modelling is pretty fast and capable. Keep in mind ive only put a few days of effort into learning the software so there is alot i could probably do to speed up even more.
Pior, personally I can only point to my own portfolio, since I have no idea what goes on in the blender community nowdays, with exception for the obvious mudbox models, blender was the main tool I used for most of that, while not the best, it's all capable art, limited only by my own skillset so far.
Hey, eld. Were you using Blender to generate all of those Bionic Commando assets (normal maps too)? I am personally interested in Blender for its potential use in game asset creation.
Hey, eld. Were you using Blender to generate all of those Bionic Commando assets (normal maps too)? I am personally interested in Blender for its potential use in game asset creation.
PS. Was sorry about the Grin closure.
When it came to using blender in the pipeline at grin back then, it worked quite well, stuff was exported from max, so I had to learn max well enough to do some work in it to, but generally I would create sub-d highpoly models in blender, a lowpoly, unwrap, and then bring it over in max to bake, I'd also do manual retopology work in blender from any sculpted stuff, it has some great functions for retopology, and aligning the lowpoly to the highpoly as you model it.
Back then I hadn't read up much on how baking worked in blender, and as for displaying normalmapped content in blender I still get a headache over the fact that it will mirror normals if you mirror the uv's, a bit of the right tool for the right job, max can push quite some amount of vertices on static meshes, while blender has to struggle a bit, became a viewport issue, but modelling anything in blender went way faster for me.
Also, smoothinggroups were a bit of a mess since max has its own way, while blender uses the more proper edge-splits that maya also uses, solved that by pre-splitting them with the modifier in blender before exporting, to preserve splits.
Now-a-days I've been working a bit with the source engine, at which I've managed to do the whole pipeline in blender, since there's working source exporters for that.
@eld,
Its awesome that you were able to use blender in a studio
I would have thought that there would have been quite a bit of resistance to that...did your leads have any thoughts on you using it?
Wasn't really a problem, as long as I learned how to use max well enough to be able to fit in the pipeline, like fixing old objects or setting stuff up for exportation when I had to, but that goes in any studio.
Since there was no cost involved, it didn't matter that much that I had an extra tool that would improve my workpace.
Most people wont have the luxury to be able to install their main app in a few seconds or convincing the cost for buying them a copy.
When it came to using blender in the pipeline at grin back then, it worked quite well, stuff was exported from max, so I had to learn max well enough to do some work in it to, but generally I would create sub-d highpoly models in blender, a lowpoly, unwrap, and then bring it over in max to bake, I'd also do manual retopology work in blender from any sculpted stuff, it has some great functions for retopology, and aligning the lowpoly to the highpoly as you model it.
Back then I hadn't read up much on how baking worked in blender, and as for displaying normalmapped content in blender I still get a headache over the fact that it will mirror normals if you mirror the uv's, a bit of the right tool for the right job, max can push quite some amount of vertices on static meshes, while blender has to struggle a bit, became a viewport issue, but modelling anything in blender went way faster for me.
Also, smoothinggroups were a bit of a mess since max has its own way, while blender uses the more proper edge-splits that maya also uses, solved that by pre-splitting them with the modifier in blender before exporting, to preserve splits.
Now-a-days I've been working a bit with the source engine, at which I've managed to do the whole pipeline in blender, since there's working source exporters for that.
Thanks for the insight into your pipeline. Blender definitely has some nifty features and I'm hoping it won't be too long before there is better support for game development tools and work-flows. It would be great to have a viable alternative to the big commercial apps, especially for freelancers or indie devs who might not be able to splash out an a Max or Maya license.
For anyone interested in a viable game development option using Blender, I would point you to BlendElf
This is a new game engine that is being developed with Blender as the targeted modeling tool. One of the biggest issues of using Blender in a commercial environment is that everyone sets their pipelines up with exporters for Max, Maya, or SoftImage. If no one bothers to develop a Blender-specific exporter, it is nearly impossible to use it directly in the pipeline. This new BlendElf engine uses Blender specifically as its primary tool for developing models, environments, and even connecting game logic. It looks like it has some pretty decent graphics features too.
Unity is also a game development system that supports Blender.
Wasn't really a problem, as long as I learned how to use max well enough to be able to fit in the pipeline, like fixing old objects or setting stuff up for exportation when I had to, but that goes in any studio.
Since there was no cost involved, it didn't matter that much that I had an extra tool that would improve my workpace.
Most people wont have the luxury to be able to install their main app in a few seconds or convincing the cost for buying them a copy.
Hey thank you so much! Opening Blender an being able to NAVIGATE THE VIEWPORT (!) is quite the feeling, its almost alien haha. Great move from whoever implemented that.
I am getting very poor 3D performance tho, I am a bit surprised. I was expected Silo/Modo smoothness and antialiasing but it turns out to be quite slow?
The user interaction really has problems. I cant believe that I cant right click an object in the lister to bring up its properties. Its ... madness haha. There is absolutely no reason for not implementing that!
Make sure that VBOs is checked in the Preferences> System tab.
That forces Blender to use vertex arrays for mesh rendering which is far more efficient than the default immediate mode rendering.
That should speed everything up by a good amount.
I made this video when VBOs were first introduced to show the improvement...quite a difference eh?
*Remember to hit "save as default" before you exit the preferences screen or the changes wont save*
It baffles me, it sounds so easy to do. With ultra popular apps like Max and Maya, the best thing the masterminds behind Blender could do to attract users is to simply give them an environment they would be familiar with. Thats one of the reason why Mudbox got successful - its authors understood that messing with the muscle memory and habits of their potential user-base wasn't worth it...
You have to understand that NaN, the original authors of Blender based the interface design around an alternative philosophy (which I can't remember the name of now it's that long ago). With that background, and whilst under NaN's control it was never going to be 'industry standard' in what anyone would reasonably consider 'normal'.
You also need to keep in mind.. the reason you don't see good quality work with Blender isn't because it doesn't exist, it's there, just buried under tonnes and tonnes of content made by amateurs and kids.. this is Blenders Achilles heal, it's entire reputation and histrionic has been built solely on the basis of uninformed and impatient children (I mean that literally BTW and not as an offhand insult to anyone) trying to get to grips with a complex task, being flippant and spreading misinformed rhetoric that the app is shit "because I couldn't get it to work".
I've been writing tutorials on and using Blender to make game content for what must be getting on ten years now and wish I had the time to answer calls to show-and-tell to the degree chaps like you need in order to be properly understand the app isn't a total pile of shat As it is, I know what it can do and compared to other applications it does those tasks just as well.
Don't be put off by the age and inexperience of the user-base, it has absolutely no bearing on how good (or not) the application actually is.
Heya Kat, I hear you. Yup I am aware of the whole history behind Blender, the non-standard approach and so on. I think my big wish is that someday, someone grabs the open source code and massages it into something simple and standard, completely remote from the main development of the 'real' Blender itself. I am thinking along the lines of making a Silo-esque app, out of the Blender core ...
Has anyone had much experience with blenders normal map baking and unreal. does it sync well?
It works ok, though you have to invert the red and blue channels for it to look correct.
TBH i hardly ever use Blender internal baking for normals as you can get better detail from xNormal, but the AO and Displacement quality are pretty sweet if you dont mind waiting for the renders
I used Blender almost exclusively for a year and I really liked the baking. Don't have to do a lot of setting up, very easy to tweak and with good results to boot. These days I prefer xNormal, though.
After using Blender for a long time, these days I've been playing around with Max. While I think some of the basic modeling tools are a lot more intuitive in Blender, creating complex high poly geometry still seems a lot more efficient in Max with access to all the different kinds of modifiers.
I think the biggest issue regarding baking in Blender is its lack of flexibility.
I dont think they could have found a faster and easier way to bake maps (2 or 3 clicks and you are done ) but it is quite restrictive in the fact that you cant use cages, average normals or flip channels before rendering so to be usable you need to edit maps in an external program.
I really love blender but it definitely needs a good amount of work to iron out its short comings.
I think the main problem is the way in which Blender is developed means that releases surge in a single direction until the current tools become a stagnant(the BGE is a perfect example of this) and those features get a revival...there is no real continuity where everything is developed simultaneously.
Everyone seems to rave about Blenders UV tools but imho i really dont think they are everything they are cracked up to be.
A few years ago they were pretty great but again, these have now fallen to being average. There is no relax tool and adding or removing edgeloops destroys any UVs that are on a mesh (even with the preserve UVs on).
Its almost impossible to get a perfect UV map without major vert pushing within the editor
To work around this i find myself editing the seams in Blender and then using Zbrush UVMaster and Unwrap 3d pro to do the final unwrap and any tidying that is needed.
The modelling tools are pretty good and can be very efficient (if you know the shortcuts) but again lack such standard tools as inset, a consistent Bevel (3 different ways?) or the ability to remove multiple edge loops at the same time.
Layers(as in Photoshop style) would be great too as the current outliner is just not versatile enough and quickly becomes cluttered as there is no way to group meshes within it.
Also the modifiers are very inconsistent in regard to the way they work or are labeled...The SimpleDeform modifier is a perfect example of this...Why doesn't it have the bend amount in degrees rather than a factor that goes 0-10 and 2 limit options that go 0-1? Just having this fixed would make all the difference when bending complicated meshes that require precession.
Though after saying all this, i do think Blender is going in the right direction with 2.5 but needs to focus much more or the core tools rather than all the extra stuff like smoke simulations and sculpting as i feel that they are taking away from Blenders strengths and thus making it a weaker program overall.
I agree with the BGE thing...now ever for non-tech savy people, unity is a better option.
I also agree about the UVing, it used to be really quick and good, but loads of other (some free) programs do it the same way, but with more customisability. I definitely feel it Blender's development looks like it's stagnating, but really I think it's just their getting settled into the whole 2.5 thing...maybe there will be more noticeable developments after they have 2.5 done and dusted.
Polycount should have a Blender only comp...to counter all that shitty Blender art you see online
I am a maya/ zbrush/ lightwave/ shiva user... When i joined up with a couple half- assed programmers i tried to get them into 3d and making iphone games, they were looking for free 3d software. For me to try to learn blender just to teach them 3d was a bad idea. Im comfortable with maya and zbrush, i can do everything i need with them. Blender was cool for many reasons, but i personally never want to see it again. It is cool because it can teach anyone about game engines and 3d. It is cool because someone who doesnt want to pay for software can open up and look at my 3d models that i make with maya/ zbrush.
I remember one of the programmers came in one day and said he had mastered blenders interface, and i laughed to myself because 'knowing the interface', he still had no idea how to model, animate, texture, or do ANYTHING useful at all with blender. I do acknowledge that blender can do anything (model, animate, texture, blah), but i think what it comes down to- is that most 3d professionals are willing to pay for software, and if you have maya zbrush or lightwave , then its quite pointless to go out of your way to learn blender for yourself.
When i was with these programmers, i was desperately trying to find a good game engine to make iphone games with. They were trying oolong, and sio2.... I left that situation thinking: if a group of people that all make money making websites arent willing to pay a little money for a game engine license, then they arent in it to win it. The best things in life cost money, period. The free game engines were crap, with no support. I needed a next-gen engine that could handle my maya rigs, blender has a totally different rigging/ animation situation, and for someone who has been rigging/ animating in maya for 5+ years, its just not worth learning at all.
Conclusion: blender is a free way of showing most people that they arent motivated enough to learn anything about 3d. If someone sees you working in 3d and say wow thats cool, it looks easy, i bet i could learn it. tell them to download blender and laugh in their face. 99 percent of the world will download it, open it, close it immediately, and then say to themselves: i have a pro 3d program now, im the man. Then never open it again.
Metalliandy, what there's uv-preservation in blender, I had no idea
ps. start pinning stuff in the uv-editor, lscm is basicly a better form of relax anyway isn't it?, usually if I have a heavy area on the mesh that I change drasticly, I would pin everything, unpin the area that is completely ruined, and unwrap, and it'll unwrap the newly modelled/changed area perfectly.
but yes, the uv-preservation, adding new edges and cutting in the mesh ends up with more trouble than it should be.
mirroring, aligning vertices and all that is very useful, average uv-islands aswell, display amount of stretching is wonderful, actually, having all the mesh editing tools in the uv editors, that kicks ass.
Actually, in the best case scenario I can pre-plan and put my seams out, hit unwrap, and basicly just hit average islands to get a unison texel density, and after that just tetris the pieces together on the map.
Metalliandy, what there's uv-preservation in blender, I had no idea
ps. start pinning stuff in the uv-editor, lscm is basicly a better form of relax anyway isn't it?, usually if I have a heavy area on the mesh that I change drasticly, I would pin everything, unpin the area that is completely ruined, and unwrap, and it'll unwrap the newly modelled/changed area perfectly.
but yes, the uv-preservation, adding new edges and cutting in the mesh ends up with more trouble than it should be.
mirroring, aligning vertices and all that is very useful, average uv-islands aswell, display amount of stretching is wonderful, actually, having all the mesh editing tools in the uv editors, that kicks ass.
Actually, in the best case scenario I can pre-plan and put my seams out, hit unwrap, and basicly just hit average islands to get a unison texel density, and after that just tetris the pieces together on the map.
well not really...i meant transform correction :P typo
I always found pinning UVs a really needless and time consuming task when many programs will do this automatically.
The algorithm that UVMaster uses is available as a white paper which could be integrated into blender but there doesn't seem to be much interest in developing it currently...i hope that changes though
The UV editor has it up point as you said...its really nice to be able to use the same tools as in the rest of the program
One thing i would love to see is an edit poly type modifier in Blender, so i could add extra loops for baking and then delete the modifier to regain the original mesh
Yea defiantly for subd but also for the LP before baking maps.
Currently if i want hard edges on the LP have to split the mesh with the EdgeSplit modifier which causes seams to show in the normal map.
If there was an Edit Poly modifier, i could add extra supporting loops to the LP negating the need (most of the time) for the pre bake EdgeSplit and thus no seams.
I could then delete the modifier and regain the original LP mesh
Of course Blender would need UV preservation first
I don't understand what you are saying there...why would you need to add loops to the LP and then take them away after the bake? Surely the mesh is going to be projecting for the final LP anyways, so it makes no difference?
Say you have a perfect LP mesh that has good UVs but when you bake you get the nasty shading errors across the faces.
The only ways to remove these errors in Blender are to split the edges with the EdgeSplit modifier, separate the faces, Bevel the edges or add extra supporting loops so that the smoothing errors are much less visible.
Splitting the faces or using the EdgeSplit modifier causes seams in the normal map wherever you split the geo and Beveling the mesh adds extra geo.
By adding extra (temporary) supporting loops on the LP you can retain the integrity of the original LP mesh but gain a much cleaner bake with no seams.
Here is a very basic example of what i mean.
The first image is the bake with just a standard cube, the second is the same cube with the EdgeSplit modifier applied and the third is with the extra loops.
The example is just to show the lack of shading errors in the bake.
I hope that makes sense
*Edit*
Added better examples
The first image is the bake with just a standard LP mesh, the second is the same mesh with the EdgeSplit modifier applied and the third is with the extra loops.
Basically what you are doing is smoothing the LP by adding extra loops, rather than splitting the verts.
There is ... absolutely nothing wrong with that cube. If your baking app produces that, and your engine or viewer is synched to it, it would work just fine. (did it just the other day) Its not a shading error, its math ...
@Pior
Yes, the gradients would be correct if i wanted a smoothed cube but im talking about hard edges...maybe i didn't explain what i was trying to achieve very well..my bad.
What i was trying to show with the cube example is how to reduce the gradients if you want hard surfaces/edges without adding smoothing groups/splitting the verts.
That mesh was baked only to show the effectiveness of removing the gradients by adding loops as the Base LP cube and the Extra Loops example both contain only 1 smoothing group (default all smooth).
btw. did you try enabling VBOs in 2.5 to improve the performance?
But that the thing, these gradients are here in order to 'balance' the smoothness of the soft edged cube. With a correct synching between baker and engine, it should be just fine, the cube with the normalmap on it should show up just fine, with nice hard edges.
Now I wouldnt work that way : I would make the edges hard on the lowpoly and separate the UVs on these edges too. But this is more for practical reasons.
If you have to jump through the hoops you describe, that means something is wrong in the pipeline...
Had no chance to try out the VBO settings, because it was slow even on a 12 triangles cube ... And the Maya navigation was here, but not the components modes or anything else .... Hence I ll just patiently wait for the next official release :P
I have to say it was quite comfortable to work with blender EXCEPT you wasn't able to set your own keys .. (talking about version 2.4) but it was possible to get used to it.
Pior is right there, blender follows the standard rules of normalmap baking, the normalmap ends up very gradient due to trying to make up for the smooth lowpoly cube, the only valid question here would be if it renders like max did in the great epic thread of max-baking.
So either it bakes right and you dont have to do much, since it will make up for the smoothing in the cube, or it will bake with the wrong tangentbasis and you'll have to start splitting edges and notify the blender authorities of that error (the one that max suffers from too).
Pior, if you use maya or max is it in opengl or d3d mode?, it sounds like blender is being forced into software mode, and yeah, hopefully all functions will be translated into hotkeys so you can use it.
... And the Maya navigation was here, but not the components modes or anything else .... Hence I ll just patiently wait for the next official release :P
... and yeah, hopefully all functions will be translated into hotkeys so you can use it.
The problem here is that Maya/3dsmax and Blender have a different set of functions and tools. While some of them behave like their counterparts in other 3d applications, many are unique in the way they work and can't be directly remapped to the same shortcuts and provide the same user interaction as you find it in other 3d applications. Design differences can be seen even at the most basic level. In 3dsmax, you first pick a tool (e.g. extrude) and then pick the mesh elements you want to use the tool on. In Blender you first select the mesh elements and then the tool. First tool -> selection vs. First selection -> tool. The rule is not universal, but it mainly stands.
As the developer that added the preset feature puts it:
this is more about giving people the ability to comfortably get around in Blender, to migrate to it, to use it in a pipeline (eg. unwrapping characters for 3ds max without interrupting workflow) or even just to evaluate it without the initial 'argh wtf!' moments getting in the way.
Blender will never be able to exactly mimic other apps, and nor should it anyway. But we can make it more comfortable for people to ease their way into Blender by removing some of the basic muscle memory roadblocks that experienced users often immediately run into.
While the extent to which Blender's UI can be shaped and customized has not yet been explored (the python defined UI is rather flexible) I don't think you will ever be able to make it behave exactly like another 3d application.
Regarding the slow rendering of the interface, do you perhaps have an ATI GPU? ATI isn't exactly known for having good OpenGL support in their drivers.
So yeah obviously I am not looking for every shortcut to be cloned from max/maya to Blender, that would be just stupid. I dont care for M bringing up the material editor...
For me its really all about the basics. I mean sure there is some differences between apps, but I am pretty certain that Blender users edit vertices, edges and faces, right? Thats really all that I am asking.
So, we have the navigation part sorted out, thats pretty impressive already. Now what would be smart, would be to have some kind of simple splash-page accessible tutorials about how to do basic stuff in blender. Selecting verts, moving them in screen mode, and so on. (Tutorials made by people knowing how to model professionals, not by kids who just think they "figured out threedee")
Pushing it further, would be a collaboration between two artists. One does a mini video showing very clearly how to do something in Max and or Maya. Then the Blender user records the same action, in Blender ways. That way everybody learns...
Another solution (yet involving more coding), would be to have some extra (non mandatory) floating UI elements doing basics things the way other programs do. Verts/edge/faces modes ; Axis constraints ; screen mode/parent mode/local mode ; a panel grouping some common polymodeling tools (extrude, cut...), and another one for basic modifiers (bend, symmetry, and so on). With the Python UI scripting thing its bound to be super easy to do ...
But that the thing, these gradients are here in order to 'balance' the smoothness of the soft edged cube. With a correct synching between baker and engine, it should be just fine, the cube with the normalmap on it should show up just fine, with nice hard edges.
Now I wouldnt work that way : I would make the edges hard on the lowpoly and separate the UVs on these edges too. But this is more for practical reasons.
Pior is right there, blender follows the standard rules of normalmap baking, the normalmap ends up very gradient due to trying to make up for the smooth lowpoly cube, the only valid question here would be if it renders like max did in the great epic thread of max-baking.
So either it bakes right and you dont have to do much, since it will make up for the smoothing in the cube, or it will bake with the wrong tangentbasis and you'll have to start splitting edges and notify the blender authorities of that error (the one that max suffers from too).
I know what you guys are getting at and of course you are both perfectly correct
The cube was a really bad example
I guess all im saying is that this technique can be used as an alternative to baking with sharp edges if needed.
Pushing it further, would be a collaboration between two artists. One does a mini video showing very clearly how to do something in Max and or Maya. Then the Blender user records the same action, in Blender ways. That way everybody learns...
I think this idea would be awesome! I would be happy record some stuff if you wanted.
What sort of thing did you have in mind?
I guess all im saying is that this technique can be used as an alternative to baking with sharp edges if needed. ?
It should not be needed though , But this is another interesting point, means blender bakes after render-tangent basis and not screen, it has been the method to fight that issue though, either splitting the edges to create a less chaosy normals, or adding support seams which you did there.
metalliandy: I really think you don't get it, though. Unless we're all misunderstanding you, ofcourse.
When you break edges, you need to break the uv's along those edges as well, and move them apart a few pixels. This is documented well on the Polycount wiki, which is in my signature. I wouldn't bake something hard-edged with one smoothing group, most of the time, for multiple reasons. For instance: If you bake with broken edges/smoothing groups/split normals, your bake won't have these gradients and manipulating your normalmap will be easier (painting out errors, for instance)
It's also more robust. A complicated mechanical shape with a lot of straight angles will not bake as nicely with one smoothing group (or in your case: an unbroken mesh. Is there no way to break the normals without breaking the edge, in Blender?).
Adding those edge you're talking about will make your bake less accurate, as you're baking to a lowpoly with different vert-normals than your final mesh. You may not have noticed the errors, but I assure you they will be there, subtle or not. While you may be able to get away with it, there are well-documented ways of working that don't have this problem.
Sorry if i have not been making much sense and using bad examples to explain myself.
I have been splitting the edges due to the same sort of problems in blender that max has but i didn't realise that adding loops in this way caused problems in the final mesh because of the different vert-normals.
I did test a while back using EQ's test mesh from the max normal map thread and got load of errors in the viewport, so i thought i would give it a try again and got the same results.
(i reported it in the Blender Bug Tracker but didnt get a response yet :S)
So it seems that Blender has the same issue, would you all agree?
So yeah obviously I am not looking for every shortcut to be cloned from max/maya to Blender, that would be just stupid. I dont care for M bringing up the material editor...
For me its really all about the basics. I mean sure there is some differences between apps, but I am pretty certain that Blender users edit vertices, edges and faces, right? Thats really all that I am asking.
Ok, just wanted to make sure people don't get the wrong idea The basics are there alright, but things can get different very early. Even with the way you access the move tool. You can do it via the manipulator, but you're potentially leaving out a lot of the extra options you get when using the shortcuts directly. And the move tool is used in the same way in other parts of Blender (UV editor, graph editor etc.) where you don't have a manipulator and access the move tool with the shortcut.
So, we have the navigation part sorted out, thats pretty impressive already. Now what would be smart, would be to have some kind of simple splash-page accessible tutorials about how to do basic stuff in blender. Selecting verts, moving them in screen mode, and so on. (Tutorials made by people knowing how to model professionals, not by kids who just think they "figured out threedee")
If the tutorials are made, it shouldn't be hard to link to the webpage. You already have a link to the official online manual, but it probably isn't enough straight to the point. In the top toolbar help -> manual.
Pushing it further, would be a collaboration between two artists. One does a mini video showing very clearly how to do something in Max and or Maya. Then the Blender user records the same action, in Blender ways. That way everybody learns...
Would require more work, but could be very useful. But also easier to make since it's very goal oriented instead of a tutorial that tries teaching you everything.
Another solution (yet involving more coding), would be to have some extra (non mandatory) floating UI elements doing basics things the way other programs do. Verts/edge/faces modes ; Axis constraints ; screen mode/parent mode/local mode ; a panel grouping some common polymodeling tools (extrude, cut...), and another one for basic modifiers (bend, symmetry, and so on). With the Python UI scripting thing its bound to be super easy to do ...
You can already do this to an extent without editing the UI scripts. in 3d window, press T (or if the shortcut doesn't work) go to the bar at the bottom, view->Tool Shelf and you'll get a side panel in the 3d window, listing all the common tools, also depending on the mode you're in (object, mesh, texture paint, sculpt etc.) You can also add your own tools.
...So, we have the navigation part sorted out, thats pretty impressive already. Now what would be smart, would be to have some kind of simple splash-page accessible tutorials about how to do basic stuff in blender. Selecting verts, moving them in screen mode, and so on. (Tutorials made by people knowing how to model professionals, not by kids who just think they "figured out threedee")
Pushing it further, would be a collaboration between two artists. One does a mini video showing very clearly how to do something in Max and or Maya. Then the Blender user records the same action, in Blender ways. That way everybody learns...
This is something I've been wanting to do for a *long* time, it's tricky though because the people you need access to to do this sort of thing are already very busy with their own projects/jobs/RL (like yourself for example), it's a challenge to say the least to pin people down for any length of time to be able to do something like this justice. But I 100% agree, it's what's needed, and not just for Blender converts - it would be an interesting series of cross platform tutorials that would allow people to go from one app to the other regardless.
...I did test a while back using EQ's test mesh from the max normal map thread and got load of errors in the viewport, so i thought i would give it a try again and got the same results.
(i reported it in the Blender Bug Tracker but didnt get a response yet :S)
So it seems that Blender has the same issue, would you all agree?
I'm struggling to understand what you're trying to do here because that looks like you've applied a uniform Set Smooth to the entire mesh, absent of *any* edge splits and then baked that. If that's the case then you'll never get the bake I think you're expecting because the process just doesn't work like that (as you know)... You'd need to use edge-splits to get a clean bake (albeit then have anti-aliasing issues to sort out). If you were using that as an in-game model, would it have a single smooth-group?
I'm struggling to understand what you're trying to do here because that looks like you've applied a uniform Set Smooth to the entire mesh, absent of *any* edge splits and then baked that. If that's the case then you'll never get the bake I think you're expecting because the process just doesn't work like that (as you know)... You'd need to use edge-splits to get a clean bake (albeit then have anti-aliasing issues to sort out). If you were using that as an in-game model, would it have a single smooth-group?
I'm struggling to understand what you're trying to do here because that looks like you've applied a uniform Set Smooth to the entire mesh, absent of *any* edge splits and then baked that. If that's the case then you'll never get the bake I think you're expecting because the process just doesn't work like that (as you know)... You'd need to use edge-splits to get a clean bake (albeit then have anti-aliasing issues to sort out). If you were using that as an in-game model, would it have a single smooth-group?
Incorrect. Needing to do any of that stuff is evidence that your baker isn't synced with the realtime shader.
Replies
As its a timelapse i cropped out the ui as it would be pretty useless to look at.
Not the best model nor am i an amazing modeller but it should show that the modelling is pretty fast and capable. Keep in mind ive only put a few days of effort into learning the software so there is alot i could probably do to speed up even more.
http://www.vimeo.com/11761088
So far the main thing that I'm liking is the softselection falloff being on the mousewheel, it makes using it a joy.
Hey, eld. Were you using Blender to generate all of those Bionic Commando assets (normal maps too)? I am personally interested in Blender for its potential use in game asset creation.
PS. Was sorry about the Grin closure.
When it came to using blender in the pipeline at grin back then, it worked quite well, stuff was exported from max, so I had to learn max well enough to do some work in it to, but generally I would create sub-d highpoly models in blender, a lowpoly, unwrap, and then bring it over in max to bake, I'd also do manual retopology work in blender from any sculpted stuff, it has some great functions for retopology, and aligning the lowpoly to the highpoly as you model it.
Back then I hadn't read up much on how baking worked in blender, and as for displaying normalmapped content in blender I still get a headache over the fact that it will mirror normals if you mirror the uv's, a bit of the right tool for the right job, max can push quite some amount of vertices on static meshes, while blender has to struggle a bit, became a viewport issue, but modelling anything in blender went way faster for me.
Also, smoothinggroups were a bit of a mess since max has its own way, while blender uses the more proper edge-splits that maya also uses, solved that by pre-splitting them with the modifier in blender before exporting, to preserve splits.
Now-a-days I've been working a bit with the source engine, at which I've managed to do the whole pipeline in blender, since there's working source exporters for that.
Its awesome that you were able to use blender in a studio
I would have thought that there would have been quite a bit of resistance to that...did your leads have any thoughts on you using it?
Since there was no cost involved, it didn't matter that much that I had an extra tool that would improve my workpace.
Most people wont have the luxury to be able to install their main app in a few seconds or convincing the cost for buying them a copy.
Thanks for the insight into your pipeline. Blender definitely has some nifty features and I'm hoping it won't be too long before there is better support for game development tools and work-flows. It would be great to have a viable alternative to the big commercial apps, especially for freelancers or indie devs who might not be able to splash out an a Max or Maya license.
This is a new game engine that is being developed with Blender as the targeted modeling tool. One of the biggest issues of using Blender in a commercial environment is that everyone sets their pipelines up with exporters for Max, Maya, or SoftImage. If no one bothers to develop a Blender-specific exporter, it is nearly impossible to use it directly in the pipeline. This new BlendElf engine uses Blender specifically as its primary tool for developing models, environments, and even connecting game logic. It looks like it has some pretty decent graphics features too.
Unity is also a game development system that supports Blender.
Awesome!
Thats good to know
I am getting very poor 3D performance tho, I am a bit surprised. I was expected Silo/Modo smoothness and antialiasing but it turns out to be quite slow?
The user interaction really has problems. I cant believe that I cant right click an object in the lister to bring up its properties. Its ... madness haha. There is absolutely no reason for not implementing that!
Make sure that VBOs is checked in the Preferences> System tab.
That forces Blender to use vertex arrays for mesh rendering which is far more efficient than the default immediate mode rendering.
That should speed everything up by a good amount.
I made this video when VBOs were first introduced to show the improvement...quite a difference eh?
*Remember to hit "save as default" before you exit the preferences screen or the changes wont save*
You also need to keep in mind.. the reason you don't see good quality work with Blender isn't because it doesn't exist, it's there, just buried under tonnes and tonnes of content made by amateurs and kids.. this is Blenders Achilles heal, it's entire reputation and histrionic has been built solely on the basis of uninformed and impatient children (I mean that literally BTW and not as an offhand insult to anyone) trying to get to grips with a complex task, being flippant and spreading misinformed rhetoric that the app is shit "because I couldn't get it to work".
I've been writing tutorials on and using Blender to make game content for what must be getting on ten years now and wish I had the time to answer calls to show-and-tell to the degree chaps like you need in order to be properly understand the app isn't a total pile of shat
Don't be put off by the age and inexperience of the user-base, it has absolutely no bearing on how good (or not) the application actually is.
TBH i hardly ever use Blender internal baking for normals as you can get better detail from xNormal, but the AO and Displacement quality are pretty sweet if you dont mind waiting for the renders
I have to say I hate that whole thing with builds though, and never download them. I wait for official releases.
After using Blender for a long time, these days I've been playing around with Max. While I think some of the basic modeling tools are a lot more intuitive in Blender, creating complex high poly geometry still seems a lot more efficient in Max with access to all the different kinds of modifiers.
I dont think they could have found a faster and easier way to bake maps (2 or 3 clicks and you are done ) but it is quite restrictive in the fact that you cant use cages, average normals or flip channels before rendering so to be usable you need to edit maps in an external program.
I really love blender but it definitely needs a good amount of work to iron out its short comings.
I think the main problem is the way in which Blender is developed means that releases surge in a single direction until the current tools become a stagnant(the BGE is a perfect example of this) and those features get a revival...there is no real continuity where everything is developed simultaneously.
Everyone seems to rave about Blenders UV tools but imho i really dont think they are everything they are cracked up to be.
A few years ago they were pretty great but again, these have now fallen to being average. There is no relax tool and adding or removing edgeloops destroys any UVs that are on a mesh (even with the preserve UVs on).
Its almost impossible to get a perfect UV map without major vert pushing within the editor
To work around this i find myself editing the seams in Blender and then using Zbrush UVMaster and Unwrap 3d pro to do the final unwrap and any tidying that is needed.
The modelling tools are pretty good and can be very efficient (if you know the shortcuts) but again lack such standard tools as inset, a consistent Bevel (3 different ways?) or the ability to remove multiple edge loops at the same time.
Layers(as in Photoshop style) would be great too as the current outliner is just not versatile enough and quickly becomes cluttered as there is no way to group meshes within it.
Also the modifiers are very inconsistent in regard to the way they work or are labeled...The SimpleDeform modifier is a perfect example of this...Why doesn't it have the bend amount in degrees rather than a factor that goes 0-10 and 2 limit options that go 0-1? Just having this fixed would make all the difference when bending complicated meshes that require precession.
Though after saying all this, i do think Blender is going in the right direction with 2.5 but needs to focus much more or the core tools rather than all the extra stuff like smoke simulations and sculpting as i feel that they are taking away from Blenders strengths and thus making it a weaker program overall.
I also agree about the UVing, it used to be really quick and good, but loads of other (some free) programs do it the same way, but with more customisability. I definitely feel it Blender's development looks like it's stagnating, but really I think it's just their getting settled into the whole 2.5 thing...maybe there will be more noticeable developments after they have 2.5 done and dusted.
Polycount should have a Blender only comp...to counter all that shitty Blender art you see online
I am a maya/ zbrush/ lightwave/ shiva user... When i joined up with a couple half- assed programmers i tried to get them into 3d and making iphone games, they were looking for free 3d software. For me to try to learn blender just to teach them 3d was a bad idea. Im comfortable with maya and zbrush, i can do everything i need with them. Blender was cool for many reasons, but i personally never want to see it again. It is cool because it can teach anyone about game engines and 3d. It is cool because someone who doesnt want to pay for software can open up and look at my 3d models that i make with maya/ zbrush.
I remember one of the programmers came in one day and said he had mastered blenders interface, and i laughed to myself because 'knowing the interface', he still had no idea how to model, animate, texture, or do ANYTHING useful at all with blender. I do acknowledge that blender can do anything (model, animate, texture, blah), but i think what it comes down to- is that most 3d professionals are willing to pay for software, and if you have maya zbrush or lightwave , then its quite pointless to go out of your way to learn blender for yourself.
When i was with these programmers, i was desperately trying to find a good game engine to make iphone games with. They were trying oolong, and sio2.... I left that situation thinking: if a group of people that all make money making websites arent willing to pay a little money for a game engine license, then they arent in it to win it. The best things in life cost money, period. The free game engines were crap, with no support. I needed a next-gen engine that could handle my maya rigs, blender has a totally different rigging/ animation situation, and for someone who has been rigging/ animating in maya for 5+ years, its just not worth learning at all.
Conclusion: blender is a free way of showing most people that they arent motivated enough to learn anything about 3d. If someone sees you working in 3d and say wow thats cool, it looks easy, i bet i could learn it. tell them to download blender and laugh in their face. 99 percent of the world will download it, open it, close it immediately, and then say to themselves: i have a pro 3d program now, im the man. Then never open it again.
ps. start pinning stuff in the uv-editor, lscm is basicly a better form of relax anyway isn't it?, usually if I have a heavy area on the mesh that I change drasticly, I would pin everything, unpin the area that is completely ruined, and unwrap, and it'll unwrap the newly modelled/changed area perfectly.
but yes, the uv-preservation, adding new edges and cutting in the mesh ends up with more trouble than it should be.
mirroring, aligning vertices and all that is very useful, average uv-islands aswell, display amount of stretching is wonderful, actually, having all the mesh editing tools in the uv editors, that kicks ass.
Actually, in the best case scenario I can pre-plan and put my seams out, hit unwrap, and basicly just hit average islands to get a unison texel density, and after that just tetris the pieces together on the map.
well not really...i meant transform correction :P typo
I always found pinning UVs a really needless and time consuming task when many programs will do this automatically.
The algorithm that UVMaster uses is available as a white paper which could be integrated into blender but there doesn't seem to be much interest in developing it currently...i hope that changes though
The UV editor has it up point as you said...its really nice to be able to use the same tools as in the rest of the program
One thing i would love to see is an edit poly type modifier in Blender, so i could add extra loops for baking and then delete the modifier to regain the original mesh
Currently if i want hard edges on the LP have to split the mesh with the EdgeSplit modifier which causes seams to show in the normal map.
If there was an Edit Poly modifier, i could add extra supporting loops to the LP negating the need (most of the time) for the pre bake EdgeSplit and thus no seams.
I could then delete the modifier and regain the original LP mesh
Of course Blender would need UV preservation first
Got a visual example or something?
The only ways to remove these errors in Blender are to split the edges with the EdgeSplit modifier, separate the faces, Bevel the edges or add extra supporting loops so that the smoothing errors are much less visible.
Splitting the faces or using the EdgeSplit modifier causes seams in the normal map wherever you split the geo and Beveling the mesh adds extra geo.
By adding extra (temporary) supporting loops on the LP you can retain the integrity of the original LP mesh but gain a much cleaner bake with no seams.
Here is a very basic example of what i mean.
The first image is the bake with just a standard cube, the second is the same cube with the EdgeSplit modifier applied and the third is with the extra loops.
The example is just to show the lack of shading errors in the bake.
I hope that makes sense
*Edit*
Added better examples
The first image is the bake with just a standard LP mesh, the second is the same mesh with the EdgeSplit modifier applied and the third is with the extra loops.
Basically what you are doing is smoothing the LP by adding extra loops, rather than splitting the verts.
Yes, the gradients would be correct if i wanted a smoothed cube but im talking about hard edges...maybe i didn't explain what i was trying to achieve very well..my bad.
What i was trying to show with the cube example is how to reduce the gradients if you want hard surfaces/edges without adding smoothing groups/splitting the verts.
That mesh was baked only to show the effectiveness of removing the gradients by adding loops as the Base LP cube and the Extra Loops example both contain only 1 smoothing group (default all smooth).
btw. did you try enabling VBOs in 2.5 to improve the performance?
Now I wouldnt work that way : I would make the edges hard on the lowpoly and separate the UVs on these edges too. But this is more for practical reasons.
If you have to jump through the hoops you describe, that means something is wrong in the pipeline...
Had no chance to try out the VBO settings, because it was slow even on a 12 triangles cube ... And the Maya navigation was here, but not the components modes or anything else .... Hence I ll just patiently wait for the next official release :P
but I've only done lowpoly modeling though...
So either it bakes right and you dont have to do much, since it will make up for the smoothing in the cube, or it will bake with the wrong tangentbasis and you'll have to start splitting edges and notify the blender authorities of that error (the one that max suffers from too).
Pior, if you use maya or max is it in opengl or d3d mode?, it sounds like blender is being forced into software mode, and yeah, hopefully all functions will be translated into hotkeys so you can use it.
As the developer that added the preset feature puts it: While the extent to which Blender's UI can be shaped and customized has not yet been explored (the python defined UI is rather flexible) I don't think you will ever be able to make it behave exactly like another 3d application.
Regarding the slow rendering of the interface, do you perhaps have an ATI GPU? ATI isn't exactly known for having good OpenGL support in their drivers.
So yeah obviously I am not looking for every shortcut to be cloned from max/maya to Blender, that would be just stupid. I dont care for M bringing up the material editor...
For me its really all about the basics. I mean sure there is some differences between apps, but I am pretty certain that Blender users edit vertices, edges and faces, right? Thats really all that I am asking.
So, we have the navigation part sorted out, thats pretty impressive already. Now what would be smart, would be to have some kind of simple splash-page accessible tutorials about how to do basic stuff in blender. Selecting verts, moving them in screen mode, and so on. (Tutorials made by people knowing how to model professionals, not by kids who just think they "figured out threedee")
Pushing it further, would be a collaboration between two artists. One does a mini video showing very clearly how to do something in Max and or Maya. Then the Blender user records the same action, in Blender ways. That way everybody learns...
Another solution (yet involving more coding), would be to have some extra (non mandatory) floating UI elements doing basics things the way other programs do. Verts/edge/faces modes ; Axis constraints ; screen mode/parent mode/local mode ; a panel grouping some common polymodeling tools (extrude, cut...), and another one for basic modifiers (bend, symmetry, and so on). With the Python UI scripting thing its bound to be super easy to do ...
I know what you guys are getting at and of course you are both perfectly correct
The cube was a really bad example
I guess all im saying is that this technique can be used as an alternative to baking with sharp edges if needed.
I think this idea would be awesome! I would be happy record some stuff if you wanted.
What sort of thing did you have in mind?
I think the toolshelf (T) on the left in 2.5 is sort of like the shelves in max and maya, but one could easily be created on the top aswell.
It should not be needed though
When you break edges, you need to break the uv's along those edges as well, and move them apart a few pixels. This is documented well on the Polycount wiki, which is in my signature. I wouldn't bake something hard-edged with one smoothing group, most of the time, for multiple reasons. For instance: If you bake with broken edges/smoothing groups/split normals, your bake won't have these gradients and manipulating your normalmap will be easier (painting out errors, for instance)
It's also more robust. A complicated mechanical shape with a lot of straight angles will not bake as nicely with one smoothing group (or in your case: an unbroken mesh. Is there no way to break the normals without breaking the edge, in Blender?).
Adding those edge you're talking about will make your bake less accurate, as you're baking to a lowpoly with different vert-normals than your final mesh. You may not have noticed the errors, but I assure you they will be there, subtle or not. While you may be able to get away with it, there are well-documented ways of working that don't have this problem.
Sorry if i have not been making much sense and using bad examples to explain myself.
I have been splitting the edges due to the same sort of problems in blender that max has but i didn't realise that adding loops in this way caused problems in the final mesh because of the different vert-normals.
I did test a while back using EQ's test mesh from the max normal map thread and got load of errors in the viewport, so i thought i would give it a try again and got the same results.
(i reported it in the Blender Bug Tracker but didnt get a response yet :S)
So it seems that Blender has the same issue, would you all agree?
If the tutorials are made, it shouldn't be hard to link to the webpage. You already have a link to the official online manual, but it probably isn't enough straight to the point. In the top toolbar help -> manual.
Would require more work, but could be very useful. But also easier to make since it's very goal oriented instead of a tutorial that tries teaching you everything.
You can already do this to an extent without editing the UI scripts. in 3d window, press T (or if the shortcut doesn't work) go to the bar at the bottom, view->Tool Shelf and you'll get a side panel in the 3d window, listing all the common tools, also depending on the mode you're in (object, mesh, texture paint, sculpt etc.) You can also add your own tools.
http://boards.polycount.net/showpost.php?p=1043317&postcount=86
I only baked it with no hard edges because that was the (seemingly) suggested method
Yea its set to 1.0
Incorrect.