She Wants the Sub-D
polycounter lvl 12
Hey everyone!
So this is going to be my hard surface + PBR journal. The goal of this thread is to be as thorough and accurate as possible with every stage of the process and get infinitely better and faster. (I have EQ to thank for that, he's been keeping with my progress so far and providing me with many resources and as much feedback as he can)
Feedback/crit would be greatly appreciated. If you think I done messed up somewhere, tell me all about it.
Edit 20/7/15 - I dropped the finished blender into the viewer a little bit ago, thought I'd drop it in the post here while I'm updating. Just click on the image below and it'll redirect.
Project A - Vintage Blender/Milkshake - Marmoset Toolbag 2
I started this a few months back now and picking it up again to finish it up. It's a bit of a random choice but it didn't matter as long as it offered enough of a challenge and had a variety of shapes (milkshake mech, anyone?)
Reference board
Progress so far - High poly to the left, low poly on the right
If anyone has any suggestions on anything so far or how to approach the texturing stage, feel free to discuss. I'm currently in the test bakes and UVing phase.
So this is going to be my hard surface + PBR journal. The goal of this thread is to be as thorough and accurate as possible with every stage of the process and get infinitely better and faster. (I have EQ to thank for that, he's been keeping with my progress so far and providing me with many resources and as much feedback as he can)
Feedback/crit would be greatly appreciated. If you think I done messed up somewhere, tell me all about it.
Edit 20/7/15 - I dropped the finished blender into the viewer a little bit ago, thought I'd drop it in the post here while I'm updating. Just click on the image below and it'll redirect.
Project A - Vintage Blender/Milkshake - Marmoset Toolbag 2
I started this a few months back now and picking it up again to finish it up. It's a bit of a random choice but it didn't matter as long as it offered enough of a challenge and had a variety of shapes (milkshake mech, anyone?)
Reference board
Progress so far - High poly to the left, low poly on the right
If anyone has any suggestions on anything so far or how to approach the texturing stage, feel free to discuss. I'm currently in the test bakes and UVing phase.
Replies
My only real worry would be some of the weird normalization in the lowpoly, i'd try flipping some edges or adjust your smoothing groups a little (like make the bit where the cups go a different group from the vertical stand, so you have a hard edge at the 90degree angle).
Also, at the bottom corner on the back you've got like a mini-bevel going on that becomes a triangle. i'd say either get rid of the triangle there, or just bevel the entire edge all the way round.
Please add some screenshots of the high poly without a wireframe, or at least use isoline display. It's hard to judge the surfaces right now.
I'm not an expert but I'd consider merging the low poly. http://www.polycount.com/forum/showthread.php?t=117017
Also nice title
At the moment the low poly you have looks almost like a decimated mesh but with poor silhouette retention and un-necessary edges, I've highlighted a few problematic places that need some tlc.
Simplest fix is to add more edges to smooth out the silhouette, it's one of the biggest offenders for making low polys look bad after baking. No one likes hard edges on places that are supposed to look smooth.
Thanks Lee!
I'll definitely be paying attention to the normalization during this phase. It hasn't come up at all with my test bakes but I haven't gone as far as applying them to the mesh yet. I'll see where I end up from there
Consider that triangle section removed!
You got it!
High poly - no wires
High poly - No sub-d levels applied
Great thread, thanks for that. There's definitely a couple of things I could merge.
Thanks for that. I'm starting to notice more and more that this is a weak point of mine where I optimize the shit out of the main shapes. A friend of mine just pointed out the same thing and said to give more to the bigger, more prominent shapes for silhouette rather than the smaller pieces.
I've not thought of doing that at all. I suspected it wouldn't bake down properly (even with projection). I'll give it a shot tonight and post results
Thanks everyone for the feedback so far!
Hard edges on the low will not effect the bake in a negitive way if set up properly (uv splits at hard edges, use cage/averaged projection mesh to avoid gaps in projection).
I wouldn't do this. It will make the model less interesting. If anything add some more geometry to that part in the lowpoly mesh as it looks over-optimized.
I wasn't talking about it effecting the bake, but the end result. Having sharp edges that effect the silhouette WILL have an effect on the final model (not enough geometry means the edges will be sharp) I'm sure you knew exactly what I meant.
Ah I thought you were referring to bake issues, well, I don't think a micro bevel there will make much difference in the silhouette, as the highpoly doesn't have a large bevel in that area.
Though she may need to add some supporting geometry on the back to avoid skewing issues with the floaters, and a bevel there will help to some degree.
And EQ, when you say a cage/averaged projection mesh, do you mean softening the normals on the cage? So like in maya, view envelope then soften all edges?
For the UV's, what are your plans for the cord? You might (depending on target engine) be able to have no UV's for the cord (which also saves verts as no splits, they can be smooth/averaged normals) and just apply a simple value for albedo/spec/gloss.
If she made the wire in max with splines, and then use the ability to turn the rendering ability on both the viewport and render (thus turning the spline into a cylinder/wire) it also has the option to produce pretty much perfect uvs... she can edit this to her liking (i.e. overlaying/tilingtexture)
Sort of. That is essentially what happens, however generally when you use a cage you don't have to manually average/soften the edges or anything.
To do this in Maya all you need to do is make sure Match Using is set to Geometry Normals in the advanced option setting in transfer maps. Surface Normals will give you gaps at hard edges.
In Max, by default baking with a cage will give you an averaged projection mesh, you have to enable "use offset" to get a non-averaged projection mesh.
Using a projection mesh/cage that isn't averaged is generally a bad idea unless you don't have any hard edges.
Thanks! It'll be by hand mostly to start with. I'd like to incorporate Ddo with one of these projects for sure.
I just splined it, straightened it and left it at the top. Will definitely be applying simple values as there's really not much to it I guess.
I've merged many small elements I had detached. Although I kept a couple detached after giving me hassle by showing up in my bakes despite my cage not clashing.
UV Map - I'll be going for a 4K texture...Maybe bring it down to 2k in the end.
I tried my best with use of space on this one and switched it around a lot.
I did many test bakes and got all kinds of colorful (literally colorful,I've seen a blue I've never seen before) results. This is my latest and it's still got a few problems I'll be addressing. Things were done from Max - > xNormal using SBM.
Shots of the NM applied
Look a little closer...
And whoop. There it is.
I can sorta guess what's going wrong with BCD but A kind of caught me off guard. It's a little wavy and I'm not sure exactly why. Thoughts?
EQs thread should answer those questions: http://www.polycount.com/forum/showthread.php?t=81154.
Basically those flat surfaces are getting projected at different angles and it is hard to get the detail on them to pick up correctly. You can fix this by having one, or a few, verts placed in the middle of those surfaces to make sure that your projection is straight onto that detail. On the back area you might need a couple edge loops to hold that surface better, but just test it out and see what happens.
The Issue in C looks like a bit depth issue. You should always bake your normals in 16bit format in xnormal using the tiff file format, and then down sample it to 8bit to render with. You will get better gradients and dithering in your normal map values.
Edit- For the C issue, make sure you have your tangent basis in Marmoset set to xnormal.
for C, as Bardler mentions you should bake in 16 bit .tiff format, then downsample in photoshop (or just resave as .psd if rendering in toolbag, and keep it at 16bit).
Also, perhaps straighten the large island at the bottom middle of your UV's, and the three vertical islands to the right of that. Looks like EQ was correct with his prediction that you'll need more geo on the back to bake the floaters too. Perhaps (since they're decently large shapes) model them in.
Also, post a shot of your lowpoly with a grid texture applied so we can see uv density.
Updates
UV Map - Changed things around to use as much space as possible. Small duplicated chunks were moved 1 unit over. I straightened out even more of the shells out, including the cylindrical top. Also did a tiling block for the wire.
Normal Map - Added a bit more geo and switched/tweaked some of the topology around for a cleaner bake. Definitely better than my last one. I went ahead and baked this in Max (8bit TGA) rather than XN which was...interesting.
Tex Density
RGBCMYK Mask for texturing
Here's my understanding of what the materials are by colour. Please feel free to correct me on any of these.
R - Steel
G - Painted Metal
B - Chrome
C - Rusty Metal
M - Plastics
Gonna start blocking out the texture from here.
Still got a bunch of minor wear+tear I'd like to do and clean up where I feel it's rough around the edges. The metal parts are not really where I want them just yet either.
I used the material values chat from Quixel on the Marmoset page as a base and tweaked them from there according to reference.
I did a lot of masking with photo sourced textures and hand painted over where I felt it was needed. Any suggestions would be cool with me at this stage. Here's a little thumb of the textures so far:
Also got my AO and cavity using Knald and dropped them into the appropriate slots within marmo. Convexity was used to generate some wear too.
I think some oily finger prints could look cool on the metal rim around the logo at the top of the blender and on the metal cup. The cup's looking a bit lifeless compared to the blender atm.
If you look from the 3/4 view (left most WIP image) the metal Y shaped attachment near the base of the blender looks a bit loose or like it's floating. Might be because of the strong black outline around it. Looks good from the front view though.
The whisk parts look a bit too dull imo. If they are being used regularly they'll constantly be getting polished from the whisking and cleaning, so they could be shinier with more horizontal scratch variation around the top of them (at the top they will be less polished from cleaning & blending). Using a curves adjustment on your roughness/gloss map could work.
Here's an image that might help explain what I mean with the metal whisks:
link
Looking forward to your next update!
Happy to say I'm done and I get to shelf this! I'm considering maybe making a small scene for it, mainly a diner counter with a proper milkshake glass done beside the blender.
Thanks to everyone that helped out, especially EQ! I look forward to starting a new hard surface prop....soon. (aka my next lifetime)
EDIT: aaaaaaaaaaaaaand texture maps!
The benefits of doing so are you get edge padding while texturing without having to add dilation afterwards (Seeing seams from no edge padding while working can be annoying, especially if you're sharing wips) . And you don't need the black/undefined area to see edges either because that's what the UV overlay is for. Currently it looks like your background is just a mid value, when it could be perfectly matching each island.
Also if you didn't for this project, I totally recommend using dDo — even if you don't want to use any of the scanned materials, dDo will save you a heap of time by taking your RGB mask and setting up your .PSD. It also makes it easy to preview various material types. Annnd dynamask is awesome.
I also feel like I should mention that you chose an excellent portfolio domain name.
You sir, are correct. I just noticed it when you pointed it out. Apparently, I did not include the high poly piece it in my last bake. Thanks for that. Fixyfix! My bad!
Hi Ben! I usually do this step at the end, when I got all my stuff done but you make a valid point in that it would be easier to just have the padding in the mask. My reasoning for doing it at the end is mostly because I feel like it doesn't take too long to make and not much of a hassle
Yeah I didn't for this project but I for sure would like to try it out for future projects.
Heheh thanks! Had to be done
Thanks and nope! A good mix of photosource, masking and handpainting.
Also +1 on taking the initiative and using PBR references like they should be used and excelling at material definition.
Only a few crits -
- Massage the power cord area where it connects to the base a little bit more so it looks like it belongs a bit better.
- I wish there was a bit more geo on the base where the curve of the stand deteriorates into angles a little.
- Wish the gloss had a bit more variation on the ceramic bits, scuffed areas where there would be stuff slamming into it a lot or places where someone would handle it etc
Super cool!
Time to revive this thread! I've been meaning to for a while but had little time to do personal work for a long time now. I'm slowly picking the pace up on things and continuing to work on my hard surface stuff. For the second project, I've gone for a gun which I don't really do much of at all.
.22 Hafdasa Zonda pistol
Reference
And this is the highpoly so far, got some quick materials applied in TB2
Been challenging in good ways so far, I've learned a lot already. Pedro Amorim helped me with some techniques and taking some creative liberties here and there, and EarthQuake has dissected quite a bit for me so, thanks guys. Crit is always welcome!
Think that frame screw should be sunk into the plastic frame though, you can see the little bevel in the plastic that lets it be countersunk in deeper than the grip.