Kevin - fantastic follow-up. Was such a great read - thanks for taking the time.
Quickie for you: was this all 3D sculpting? What're we looking at exactly in terms of how something like this was made? And - where's the in-game meshes!?
Thank you Kevin for taking the time to answer those questions. I really appreciated it. The work is very inspiring to see. Hopefully my future work can come close to what you do with some effort.
-DN-: Thanks for the tip on the spacing tool. I hadn't used that before. I have used Advanced Painter for bolt/screw head placement and other things before. That tool is great though and solves some issues I was having.
Used in combination with the poly edge extruded object modeling and shell modifier, the spacing tool achieves what I was looking for in placement along a surface. So the combination is a solution to the distortion issue. Just get my contour spline to follow from the create shape from selection after selecting the edge loop contour on the underlying mesh. Good to go.
Hope to see more stuff if you have more you can show Kevin. Seeing some of the low poly in game stuff for these would be interesting as well. Thanks again for all the insight.
Adam: That was all modelled, I made the forms in max, then took some time to equally
subdivide everything into the same size of quads, exported to Zbrush , doodled then
used a photooverlay as an alpha, then as a Stencil, then doodled, split everything into
polygroups under half a million polys, exported each of those, ran a Batch optimize
with polycruncher to make it manageable, imported and then built a low poly from
the initial PRE Zbrush meshs and processed.
I'm not going to show lowpolys, certainly not right now while I'm still plowing through
the renders, I might show them later, but its not a priority to me when people can see
that stuff ingame.
Likewise I'm not going to show texturing because that would be stealing Maury's thunder,
I'm hopeful he may show some of the amazing work he did on UT3 later on when he has
time.
Another Necris piece, designed to mix and match with the archwall
Thank you for the very inspiring posts, Kevin. You blew my entire weekend by making me want to model high detail stuff like this instead of doing the nothing I was planning on doing.
Da-dang-dang-dang ! Inspiring stuff, Kevin. Aaaah, I fondy remember my student days at VFS studying your textures and models from your Rorshach period :P
I jumped right into the editor to see the in game stuff and I was amazed at how many modular pieces you get out of one part. For example, the BigNose has a small varation on the circular part. But it then gets divided up into 3 different sections for additional parts/deco, same with the YBeam.
A quick question if you get the time, do you create these objects knowing they will be split up? Do you divide them up before or after the lows are created and is it your job to divide them up or does the LD look at the asset and say "hey this part would be cool over here"? Then they just instance a section of the model and import it as a new asset?
Wow, totally blown away by this work Kevin. Really not sure how you went about some of this, and I agree with Space that there needs to be a Gnomon video from you.
I always plan to split things up beforehand a little and as I work it becomes clear, that each
design is composed of 2 or 3 base elements, so to understand how the object
works I finish out those elements more fully than the concept, ie. I build stuff that cant
be seen when the elements are combined, but will be seen when they are split.
When you are learning to Hipoly you quickly realize that the more you seperate elements
into seperate meshs, the more easily you can make the shapes without slow down from the
polygon count or restrictive edgeloops from one form interfering with another.
If while building you see the individual elements are interesting and useful, it stands to
reason that the level designers might think the same and be glad of having more control of
those 1 or 2 elements within the whole design, especially when so many of the designs dont
work if used too often.
So partly, modularity in general, is about empowering your level designers and not being so
much of an artist that you refuse to surrender control over the final design.
It's rare that I'm asked to do this, it was something I had to spend a long time 'selling' to
my superiors by putting a lot of hours outside of my schedule to gradually prove my way
of working was valid and useful. I worked this way before I worked at Epic.
I mean, who gets excited about a rectangle? Artists don't find them interesting, level
designers do however find them VERY interesting because they are the core of all
constructions.
Think of it this way, a thousand years in the future we will still need I beams and bricks
to be the underlying base layer of construction, so to make a good level you need to
be aware of this.
This is just the way my mind works, I reduce all things down to their base forms, I simplify
things so I can deal with them. It's a useful thing to an environment artist to understand
that they are simply making very fancy lego bricks.
A Playmobil castle's instruction manual is actually a very good level design document if you
have the eyes to see it for what it is. I've joked in the office that such things should be a
core component of a level designers training
My motivation for learning to work this way was simply born out of frustration. For the first
year of UT3 it was just me and Maury on the environment side of development and taking
a week or 2 to create an asset that was used once just didn't seem like the most valuable
use of my time, I couldn't see how we would meet the games deadline working that way.
Last but not least, when we moved to UE3 we lost the use of BSP’s ease in shelling out levels
and being able to change the material or repeat regularity of the texture and I felt that
modularity could be the new BSP, like Pipes are the Crates of Next Gen games.
Kawe: Usually cinema and largely my time spent travelling when I was younger was in Europe
and I periodically find I have built something I saw but forgot. I also spend a lot of time
at the prototype stage of each theme researching another cultures history and style of
construction, base relief patterns and so on, it helps a lot just to look through a book and
soak it up. You may not remember it exactly, I usually dont, but it does sink in more than we
are aware of.
YOur work is truly amazing. I have just graduated from uni and was thinking of specialising in environment but thought I would have to look more into Level design. But after seeing your work I think I am going to crack on a learn ZBrush more to get up to similar standards of your work.
Great explanation, this thread has really opened my eyes to a lot of things in the construction of UT3. After being nearly overwhelmed with the editor its great to hear some of the breakdown of workflow it suddenly makes this way more approachable.
Sidenote, it also really helps explains the naming conventions too, objects seem to be organized much better than I 'org' had thought once I looked at the modularity of the assets!
If you do get a chance I'd love to hear your thoughts on this. I really appreciate you taking the time to do what you've already done.
When looking at the plaster wall assets it seems as if you went in and modeled the stone/brick patterns out with just base rectangular shapes (mainly in the upper and lower trim elements) and then dinged them up in Z Brush but then in the plaster wall it appears it would've been just an alpha or combination of alphas made into a normal/displacement map from a photo sourced texture and pulled out in Z Brush.
Do you find you get better results from one over the other and what is a good work flow for creating that kind of asset? I've been experimenting and thought maybe modeling the individual brick/stone/blocks would be better because then you could cannibalize the first assets individual brick/stone/blocks to compose new different shaped wall assets or damaged wall assets faster. I just was wondering how you approach this after seeing these last few assets.
shitttt, I wouldnt mind having some of my character near thoses kind of friken high tech detail shit! ahah, GoodJob man, its realy nice to see you post again!
You have nice technical modeling skill, but that still make me interested since it work well alltogether
I was just playing UT3 2 days ago and as usual I was absolutely blown away by the quality of the game environments, I was playing the level with the babelspire(sanctuary team deathmatch?) and noticed how elements of the high poly had been used around the level in pieces and I was just thinking wow these people are genius... and now I know who one of these geniuses is ! haha awesome work Kevin.
Telekinetic: In the case of the ceramic wall I find I get better results by seperating the elements
such as the ceramic bricks from the sandy grout inbetween them because that allows me more
control over how worn the bricks are because I can literally wear the whole brick away by
pushing or cracking it until its positioned behind the grout.
Working this way also allows me to assign different colours to the elements and then when
I process the HI to the LOW I can include a diffuse colour information texture along with
the normal map and this saves the texture artist who moves on to it afterwards to be able
to mask select all the elements and seperate them in photoshop so they can use different
photo overlays for each material without having do it by hand.
Another advantage is that then all the Hipoly chunks are reusable for variants as I can then
reposition them. It also just makes it easier to seperate elements into subtools in Zbrush
and makes for smaller chunks of under half a million polys that polycruncher can handle.
With the plaster piece, knowing ahead of time that it would be a tiling BSP texture and all/
one colour / material I didn't seperate things out.
Your point about reuseability is absolutly the the right way to go, why model 20 bricks when
5 will do as they can be rotated into different positions to hide the fact that there are
fewer unique ones, its especially easy to hide this when the grout underneath is seperate
and there for you to push some bricks angled and further recessed into.
I've also found that with organic surfaces you are best to not use smoothing groups, it may
look a little rough and noisy when you process, but by the time you put it through the
engines compression algorithm, add a detail normal, half the texture size and then display
it on a TV in someones home, that sharpness serves you really well.
awesome thread i must say. keep it up Kevin! pleasure to see your work and i think i say for all of us poly counters when i say that we hope to see tons more along with your strong posts answering many questions we have. Epiphany moments i think they call them? any who keep it up and thank you.
Awesome to see you pimping some new work Kevin. Always nice to see the awesome stuff that comes out of Epic and the awesome art team you guys have there. I would def. say that you have gotten a firm grasp on this modeling stuff.
Can't wait to see more. Perhaps after Gears 2 ships.
Wow, awesome work, Kev. I think I enjoy seeing these pieces more on their own, sans texture. So much of that detail gets lost when placed against the rest of the scene.
I just wanted to say it's great to see you post time estimates on your work. It really helps to put things in perspective (and make me realize I'm just never given that much time for hi-rez during production ).
Lee: if you look through the posts i listed times.
I used splines set to render in viewport, then converted to editable poly and modified them,
sometimes id be lazy and duplicate old ones and FFD then roughly into place. The wires
are a big time sync to do.
Amazing stuff, I never understood how Epic and a handful of other companies could afford to do stuff like this for environment art while the rest of use are stuck using small tiling texture everywhere... guess it's all about what's going on in the game, but one day it'd be great to do stuff like this.
Question:
We try to work smart and make a number of modular pieces first that can do the job that BSP
used to do in UE2 while also defining a base architecture too. That leaves more time to focus
on the more elaborate pieces which we also tried to get 4 or 5 lowpoly variants out of for
better texture efficiency.
To clarify this: You make one High poly model and then get one normal map out of this that is used for 4-5 low poly variations of this?
I'm trying to understand the possibilities of this since it the hipoly model baked out into one normal map with a unique set of uvs doesn't seem like it would offer itself to variations... other then splitting the sections up (a base, mid, top or something like that)
Gorgeous work! I love the design and detail specially on the necro pedestal. To bad it is "flattened" to a couple of hundred triangles because of today's hardware. Very cool of you to share your work and workflow with us!
Hi Kevin, this is great work and your explanations are very handy.
Could you clarify something about your overall workflow? You say that you make modular architectural pieces for the LD's to use... Do these pieces replace BSP? Or are they used in stead of BSP? Does this mean the level designers have to wait until you have those pieces made to create the level? or do they still rough things out in BSP?
I'll post some example low poly pieces to demonstrate how many modular pieces
I can create from 1 texture at the weekend.
LD's shell out levels in bsp mostly to start, then when they are given a 'meshing' and lighting
pass the BSP is replaced by the modular walls and more decorative pieces.
Thanks for the praise, no time to post new stuff through the week, but at the weekend
they'll be new eyecandy
amazing work
I was stunned when I opened the first page of this thread and continued to be, its truly inspirational, definitely something to aspire to for me.
edit -- i just realized i haven't read this thread in detail. i'll do that .. so please ignore this question if it's been answered.
Kevin, this stuff is absolutely incredible. i've got nothing but respect for the amazing level of quality you've pulled off here, but some of it begs the question... can't a lot of these little details be handled in photoshop? why are you modeling so much stuff into the geometry?
Replies
Quickie for you: was this all 3D sculpting? What're we looking at exactly in terms of how something like this was made? And - where's the in-game meshes!?
-DN-: Thanks for the tip on the spacing tool. I hadn't used that before. I have used Advanced Painter for bolt/screw head placement and other things before. That tool is great though and solves some issues I was having.
Used in combination with the poly edge extruded object modeling and shell modifier, the spacing tool achieves what I was looking for in placement along a surface. So the combination is a solution to the distortion issue. Just get my contour spline to follow from the create shape from selection after selecting the edge loop contour on the underlying mesh. Good to go.
Hope to see more stuff if you have more you can show Kevin. Seeing some of the low poly in game stuff for these would be interesting as well. Thanks again for all the insight.
subdivide everything into the same size of quads, exported to Zbrush , doodled then
used a photooverlay as an alpha, then as a Stencil, then doodled, split everything into
polygroups under half a million polys, exported each of those, ran a Batch optimize
with polycruncher to make it manageable, imported and then built a low poly from
the initial PRE Zbrush meshs and processed.
I'm not going to show lowpolys, certainly not right now while I'm still plowing through
the renders, I might show them later, but its not a priority to me when people can see
that stuff ingame.
Likewise I'm not going to show texturing because that would be stealing Maury's thunder,
I'm hopeful he may show some of the amazing work he did on UT3 later on when he has
time.
Another Necris piece, designed to mix and match with the archwall
CeramicWall1
I jumped right into the editor to see the in game stuff and I was amazed at how many modular pieces you get out of one part. For example, the BigNose has a small varation on the circular part. But it then gets divided up into 3 different sections for additional parts/deco, same with the YBeam.
A quick question if you get the time, do you create these objects knowing they will be split up? Do you divide them up before or after the lows are created and is it your job to divide them up or does the LD look at the asset and say "hey this part would be cool over here"? Then they just instance a section of the model and import it as a new asset?
what I'm wondering is where you get the inspiration for all this.
I've been looking at the UT3 editor quite a bit but seeing the high poly versions makes you realize how much work was put into it.
really.. I've been awe-struck for the last ten minutes now and I gotta close my browser now or I'll sit her all day drooling.
Gnomonolgy video tutorial?
Spark
I always plan to split things up beforehand a little and as I work it becomes clear, that each
design is composed of 2 or 3 base elements, so to understand how the object
works I finish out those elements more fully than the concept, ie. I build stuff that cant
be seen when the elements are combined, but will be seen when they are split.
When you are learning to Hipoly you quickly realize that the more you seperate elements
into seperate meshs, the more easily you can make the shapes without slow down from the
polygon count or restrictive edgeloops from one form interfering with another.
If while building you see the individual elements are interesting and useful, it stands to
reason that the level designers might think the same and be glad of having more control of
those 1 or 2 elements within the whole design, especially when so many of the designs dont
work if used too often.
So partly, modularity in general, is about empowering your level designers and not being so
much of an artist that you refuse to surrender control over the final design.
It's rare that I'm asked to do this, it was something I had to spend a long time 'selling' to
my superiors by putting a lot of hours outside of my schedule to gradually prove my way
of working was valid and useful. I worked this way before I worked at Epic.
I mean, who gets excited about a rectangle? Artists don't find them interesting, level
designers do however find them VERY interesting because they are the core of all
constructions.
Think of it this way, a thousand years in the future we will still need I beams and bricks
to be the underlying base layer of construction, so to make a good level you need to
be aware of this.
This is just the way my mind works, I reduce all things down to their base forms, I simplify
things so I can deal with them. It's a useful thing to an environment artist to understand
that they are simply making very fancy lego bricks.
A Playmobil castle's instruction manual is actually a very good level design document if you
have the eyes to see it for what it is. I've joked in the office that such things should be a
core component of a level designers training
My motivation for learning to work this way was simply born out of frustration. For the first
year of UT3 it was just me and Maury on the environment side of development and taking
a week or 2 to create an asset that was used once just didn't seem like the most valuable
use of my time, I couldn't see how we would meet the games deadline working that way.
Last but not least, when we moved to UE3 we lost the use of BSP’s ease in shelling out levels
and being able to change the material or repeat regularity of the texture and I felt that
modularity could be the new BSP, like Pipes are the Crates of Next Gen games.
Kawe: Usually cinema and largely my time spent travelling when I was younger was in Europe
and I periodically find I have built something I saw but forgot. I also spend a lot of time
at the prototype stage of each theme researching another cultures history and style of
construction, base relief patterns and so on, it helps a lot just to look through a book and
soak it up. You may not remember it exactly, I usually dont, but it does sink in more than we
are aware of.
Couple more Necris walls -
CeramicWall2
PlasterWall
Liandri piece next, a good example of modularity
Orgsupport -
Thank you for being such and inspiration..
Daz
Sidenote, it also really helps explains the naming conventions too, objects seem to be organized much better than I 'org' had thought once I looked at the modularity of the assets!
Really fantastic stuff.
When looking at the plaster wall assets it seems as if you went in and modeled the stone/brick patterns out with just base rectangular shapes (mainly in the upper and lower trim elements) and then dinged them up in Z Brush but then in the plaster wall it appears it would've been just an alpha or combination of alphas made into a normal/displacement map from a photo sourced texture and pulled out in Z Brush.
Do you find you get better results from one over the other and what is a good work flow for creating that kind of asset? I've been experimenting and thought maybe modeling the individual brick/stone/blocks would be better because then you could cannibalize the first assets individual brick/stone/blocks to compose new different shaped wall assets or damaged wall assets faster. I just was wondering how you approach this after seeing these last few assets.
Thank you again this is really incredible stuff.
You have nice technical modeling skill, but that still make me interested since it work well alltogether
I dig
ben
such as the ceramic bricks from the sandy grout inbetween them because that allows me more
control over how worn the bricks are because I can literally wear the whole brick away by
pushing or cracking it until its positioned behind the grout.
Working this way also allows me to assign different colours to the elements and then when
I process the HI to the LOW I can include a diffuse colour information texture along with
the normal map and this saves the texture artist who moves on to it afterwards to be able
to mask select all the elements and seperate them in photoshop so they can use different
photo overlays for each material without having do it by hand.
Another advantage is that then all the Hipoly chunks are reusable for variants as I can then
reposition them. It also just makes it easier to seperate elements into subtools in Zbrush
and makes for smaller chunks of under half a million polys that polycruncher can handle.
With the plaster piece, knowing ahead of time that it would be a tiling BSP texture and all/
one colour / material I didn't seperate things out.
Your point about reuseability is absolutly the the right way to go, why model 20 bricks when
5 will do as they can be rotated into different positions to hide the fact that there are
fewer unique ones, its especially easy to hide this when the grout underneath is seperate
and there for you to push some bricks angled and further recessed into.
I've also found that with organic surfaces you are best to not use smoothing groups, it may
look a little rough and noisy when you process, but by the time you put it through the
engines compression algorithm, add a detail normal, half the texture size and then display
it on a TV in someones home, that sharpness serves you really well.
Can't wait to see more. Perhaps after Gears 2 ships.
I just wanted to say it's great to see you post time estimates on your work. It really helps to put things in perspective (and make me realize I'm just never given that much time for hi-rez during production
Great work and thanks for the tips / scripts.
-Tyler
- About whats the time frame you are given for each HP art piece?
- when making the wires did you just use the spine in Max, or do you use a script to generate them?
I used splines set to render in viewport, then converted to editable poly and modified them,
sometimes id be lazy and duplicate old ones and FFD then roughly into place. The wires
are a big time sync to do.
Question:
We try to work smart and make a number of modular pieces first that can do the job that BSP
used to do in UE2 while also defining a base architecture too. That leaves more time to focus
on the more elaborate pieces which we also tried to get 4 or 5 lowpoly variants out of for
better texture efficiency.
To clarify this: You make one High poly model and then get one normal map out of this that is used for 4-5 low poly variations of this?
I'm trying to understand the possibilities of this since it the hipoly model baked out into one normal map with a unique set of uvs doesn't seem like it would offer itself to variations... other then splitting the sections up (a base, mid, top or something like that)
The necris themed stuff is just immense.
POST YOUR SHIT MOOSE
- BoBo
Could you clarify something about your overall workflow? You say that you make modular architectural pieces for the LD's to use... Do these pieces replace BSP? Or are they used in stead of BSP? Does this mean the level designers have to wait until you have those pieces made to create the level? or do they still rough things out in BSP?
I can create from 1 texture at the weekend.
LD's shell out levels in bsp mostly to start, then when they are given a 'meshing' and lighting
pass the BSP is replaced by the modular walls and more decorative pieces.
Thanks for the praise, no time to post new stuff through the week, but at the weekend
they'll be new eyecandy
Any chance we could see the in-game stuff as well?
Your stuff is really inspiring.
I was stunned when I opened the first page of this thread and continued to be, its truly inspirational, definitely something to aspire to for me.
Kevin, this stuff is absolutely incredible. i've got nothing but respect for the amazing level of quality you've pulled off here, but some of it begs the question... can't a lot of these little details be handled in photoshop? why are you modeling so much stuff into the geometry?