Best Of
Re: CAHDD™ (Computer Aided Human Designed & Developed) — voluntary workflow transparency + watermark too
The site does not get to the point, right away. In my opinion, you need to explain exactly what is being proposed, and do it "above the fold" which is immediately on the first homepage view, before someone has to scroll down.TheAllusionist said:
I really don't understand why a lot of comments are on my writing, but it appears to be the price for being long winded.
I don't see how the CAHDD system actually works, until I get to the How it Works page, and even then I have to scroll down 1/3 of the page before I get an explanation of what the watermark actually means. This should be the first thing we see when we go to your homepage.
The problem is not the style of the images, the problem for me is the fact that all of the images on your homepage look like they are 100% AI generated. They have all the hallmarks of AI-gen content, and browsing your site there are a ton of these generated images.TheAllusionist said:
Thanks for your input and follow up, the color scheme of warms with diffuse glows for hero header and footer images is a personal choice of pursuing warm earthy tones, might not be everyone's cup of tea so I will take it under advisement. I think this groups comments definitely point to replacing the three medium images in a horizontal row on the front page. The gallery was an effort to show a sample of images at different stages labled appropriately to show the concept.
None of them use your watermark system, except the car.
The car image has your Stage 3 CAHDD watermark, but it's not a trustworthy assessment since the whole image looks like it came straight out of the generator. What human input was used here, other than prompts?TheAllusionist said:
The Porsche 944 Turbo (951) has 968 wheels (not hubcaps) on it, a popular upgrade for many late offset 944s, I myself have early offset and prefer "Phone Dials", but in the Porsche world those wheels would be recognized for what they are and not questioned. The last two sentences may come across as defensive or worse, but that is not what is intended, I just wanted to point out some conclusions being made aren't true.
When I mentioned the hubcaps, I should have explained why it looks AI generated. Look at how distorted this is, is this a 3d model? Is this an overpaint? Looks like typical AI hallucination & distortion.
Also the foglights are pretty obviously not what's in the design of the real car. The Porsche emblem, the flaring, it's all over. What exactly was your contribution to your homepage images?
The errors are everywhere. What is up with these hands?
How many knuckles?!
For me the biggest problem with the whole CAHDD concept is, it's a watermark that doesn't make sense on its own. I have to look it up to figure out it's Stage 3, and then parse what that means, and examine the image to see if I trust the assessment. Most people aren't going to take the time to do this, I only did this because I'm trying to put in a good-faith assessment of what you're trying to do.
Despite the sense you have that all the feedback is negative, it's in fact all constructive criticism, it's meant to help you improve your work. That's a foundational aspect of how the Polycount Forum works. People are generally honest, and get right to the point. And when they don't, when it devolves into ad-hominem attacks or whatever, we try to steer people back on track. The feedback is meant to help the OP improve their work.
Why not just include some text above or below the image that explains what you did? No need to search around to find an explanation, it's all right there. This is a common strategy when presenting work that was done by a team. You explain what your part was... "I modeled the interior, the exterior and scene were created by others" or even better you provide your colleagues' names to credit their contribution, while explaining exactly what your personal contribution was. Here's an example, see the sidebar: https://www.artstation.com/artwork/JeKdEd
Anyhow, I think you could salvage this initiative by coming up with some good solid terminology people could use for the credit text. And demonstrate this process with each image you have on your site.
Eric Chadwick
3 ·
Photoshop - PixelB Game Maps and PixelB Normal Tools - extensions for painting game textures
I want to share the Photoshop extensions I released recently.
I realized while working on Overwatch that ever since Quixel Suite died off, there hasn’t been a good way to paint normals or generate normal/AO/cavity maps inside of Photoshop. Blizzard had their own tools for doing so but there hasn’t been a publicly available alternative to nDo or CrazyBump for years now. Adobe retired the Photoshop 3D tools in CC ’22, and for their part seem happy to let users pay the extra $25/mo for Substance suite.
Still, Substance Designer can’t match Photoshop for speed and directness when making normal maps for trimsheets or simple props. I’d hate for the current generation of artists to miss out on the old ways of making game maps, so I wrote a pair of extensions to close the gap. PixelB Game Maps is for generating AO/cavity/normals and PixelB Normal Tools is for painting normal maps.
Both are available on Gumroad:
PixelB Game Maps: https://7134563338877.gumroad.com/l/xjdsaw
features: https://www.pixelbutterfly.com/tools/pixelb_game_maps
PixelB Normal Tools: https://7134563338877.gumroad.com/l/chfail
features: https://www.pixelbutterfly.com/tools/pixelb_normal_tools
Here’s a timelapse showing how you can use the panels to make a trimsheet:
https://www.youtube.com/watch?v=FFwNlSHJQQA
And here’s a longer tutorial on how to use both extensions to make a simple chair prop:
https://www.youtube.com/watch?v=W22Kg4GNEEY
pixelb
4 ·
Re: How The F*#% Do I Model This? - Reply for help with specific shapes - (Post attempt before asking)
polska_krowa Welcome to Polycount. Consider checking out the forum information and introduction thread.
Here's a few links to write-ups that cover how to attach similar shapes to curved surfaces:
It looks like you have a solid start with the block out and topology routing. How much geometry this model needs depends entirely on what it will be used for.
If it's for up close product rendering or VFX work then the model might need to be fairly dense to ensure consistent surface quality when rendering with complex reflection and refraction shaders. In that case the mesh density really just depends on how accurate and sharp the shapes need to be.
If it's for up close product rendering or VFX work then the model might need to be fairly dense to ensure consistent surface quality when rendering with complex reflection and refraction shaders. In that case the mesh density really just depends on how accurate and sharp the shapes need to be.
If it's for a typical game art high to low baking workflow then the mesh could probably be simplified a bit to make modeling easier and quicker. Also, depending on texture resolution limits, it may be necessary to make the high poly a bit softer or taper some of the shapes. That way all the edge highlights are still visible at normal view distances.
Here's a link to another write-up that covers just how little geometry is required to get clean high poly models:
As far as connecting all the shapes: it really just comes down to lining up the segments and routing the topology around the shape intersections. With what you already have the next step would be to establish the outer control loops around the base of the intersecting shapes then simplify the rest of the mesh by dissolving any extraneous edges.
https://polycount.com/discussion/comment/2803910/#Comment_2803910
https://polycount.com/discussion/comment/2803912/#Comment_2803912
https://polycount.com/discussion/comment/2774628#Comment_2774628
https://polycount.com/discussion/comment/2803912/#Comment_2803912
https://polycount.com/discussion/comment/2774628#Comment_2774628
As far as simplifying the high poly goes, here's one example of what that could look like for use in game art:
FrankPolygon
3 ·
Re: CAHDD™ (Computer Aided Human Designed & Developed) — voluntary workflow transparency + watermark too
To be honest, my first impression is rather negative and leaves me slightly suspicious. As inevitable some of this might seem and while it might come from a good place, this kind of feels like an attempt to normalize AI for artists who might fully reject or boycott it otherwise (whether that's sustainable is a different story).
The whole website seems to rely on AI heavily. Basically all the images seem to be AI (ironically not using your own system to declare it, as far as I can see) and it's reasonable to assume the same for the writing.
There are a lots of words, but nothing in the "about" section really tells me who is behind all this. I guess I could google your name, and I can understand the notion that this is not about you or your associates (if there are any) but I guess you can also see how that looks a bit strange. So if you are saying "This is not a product", I'm immediately asking "Is it, though"?
(Edit 2: E.g. there are registered trademarks and the icon is not in the public domain)
Again, there's lots of text. That sort of makes sense if you want to get people on board and clear open questions, but I'm not reading through all of that, especially if there's a good chance that a lot of it is AI fluff.
The system/watermark itself sort of makes sense as well, but feels way to convoluted. Very few people will keep all of those steps in mind or look it up constantly. E.g. a small AI is clear and self-explanatory, while a pictogramm of a human where the indicator gets fuller the more AI is used is a bit unfortunate.
I can see something like that being used/enforced on a platform basis, but personally, I'd rather see less watermarks on images than even more. And obviously, this would never make it to an end consumer facing product, it's more for galeries and stock and portfolios.
It might make sense as a general tool to quickly inform people about your techniques, but in its current form it seems just too convoluted, and then there's the question if you can really call AI a "technique" that has a place in a continous spectrum of art techniques, which this kind of implies.
Edited the last paragraph.
The whole website seems to rely on AI heavily. Basically all the images seem to be AI (ironically not using your own system to declare it, as far as I can see) and it's reasonable to assume the same for the writing.
There are a lots of words, but nothing in the "about" section really tells me who is behind all this. I guess I could google your name, and I can understand the notion that this is not about you or your associates (if there are any) but I guess you can also see how that looks a bit strange. So if you are saying "This is not a product", I'm immediately asking "Is it, though"?
(Edit 2: E.g. there are registered trademarks and the icon is not in the public domain)
Again, there's lots of text. That sort of makes sense if you want to get people on board and clear open questions, but I'm not reading through all of that, especially if there's a good chance that a lot of it is AI fluff.
The system/watermark itself sort of makes sense as well, but feels way to convoluted. Very few people will keep all of those steps in mind or look it up constantly. E.g. a small AI is clear and self-explanatory, while a pictogramm of a human where the indicator gets fuller the more AI is used is a bit unfortunate.
I can see something like that being used/enforced on a platform basis, but personally, I'd rather see less watermarks on images than even more. And obviously, this would never make it to an end consumer facing product, it's more for galeries and stock and portfolios.
It might make sense as a general tool to quickly inform people about your techniques, but in its current form it seems just too convoluted, and then there's the question if you can really call AI a "technique" that has a place in a continous spectrum of art techniques, which this kind of implies.
Edited the last paragraph.
Noren
3 ·
Re: CAHDD™ (Computer Aided Human Designed & Developed) — voluntary workflow transparency + watermark too
This sounds like a nice idea, but your website is coated in AI slop (How many fingers does a toddler have lol?) and it's a ton of words without actually getting to the point (where is the system actually documented?).
The whole site seems to be 90% AI created, where's the watermark explaining that?
Also, there seem to be zero human names on that website, if I go to About Us or Who We Are, there are zero people to learn about.
Sorry, I can see the idea is interesting. But the presentation seems quite tone-deaf.
The whole site seems to be 90% AI created, where's the watermark explaining that?
Also, there seem to be zero human names on that website, if I go to About Us or Who We Are, there are zero people to learn about.
Sorry, I can see the idea is interesting. But the presentation seems quite tone-deaf.
Eric Chadwick
7 ·
Re: What made you feel good today?
I've been fattening up the fluffy livestock I keep in the event of social order collapse and WW3, with slow-cooked water buffalo, rice and red cabbage. I'm pleased to announce that the small one has a decent amount of meat on his drumsticks
zetheros
4 ·
"Streamside AR8" - fictional vintage race car (with Unity webplayer)
Hi ya good people of polycount!
Thought I'd start a thread now that I got this project done. I'm really bad at documenting my progress as I go but I'll try to include the most important milestones here. I'll kick off with a few 3DO renders.
The initial idea was to just model an interesting object and then call it done. I got inspired by a photo of a race car fitted with a radial engine so I went with that. The reference image is lower down on this post. This is the engine at 600k tris, ibl clay material and rendered in mental ray.
After that, I got the idea to try making a game model "hero asset" and try normal map baking. I hadn't done any of that before so this turned into a learning project as well. And then the whole project started expanding.
Since I was already going ahead with all the normal maps I thought I'd add some more detail using nDo2 which I had used a bit at work. I'm actually quite proud on how the engine front cover engraving turned out.
Next up was trying out the Quixel DDO trial on texturing the engine. Such a fun way of working and I definitely left me wanting to try it out on another project later on!
It took friggin' ages to the normal map bakes on the first parts I did. Engine ended up having 15k tris after the normal map bakes.
[IMG]https://fc04.deviantart.net/fs71/f/2014/171/8/3/radial_8___game_model_by_ollite20- d7n6zeo.jpg[/IMG]
Like I mentioned earlier the project started expanding and just having an engine didn't feel enough so I decided to create something around it and started gathering interesting reference photos. I liked the aesthetics of the 1930's grand prix racers with the exposed steering, suspension and wheels so I went with that as the main theme.
I had had the main inspiration photo for the engine on my hard drive for a long while before starting on modeling the engine. (The car is the 1935 Monaco Trossi which I later found out was horrible to drive because of the brutal front-heavy weight distribution.)
Everything is arranged onto one texture sheet and textured in Quixel Suite. Not fully satisfied with some of the materials in there but that'll do. The project file somehow got corrupted at some point so the only way to edit the materials would be to do them all over again. Textures' size is 4096x4096.
Most important ones. There's AO and opacity too but they're not that interesting to look at so I'll leave those out for now.
Just the mesh (3D0):
Mesh + normal map (3D0):
Mesh + normals + spec + gloss (3D0):
Everything (3D0):
With UVs (3D0):
It's been a long but fun project and I've learned a lot while working on this. Many firsts in this for me. The seat leather wrinkles and the cloth wrapped around the exhaust within the engine bay, among other things mentioned, are my first try at sculpting as well. Sure, there's still a few things I'm not 100% satisfied but I gotta move on. I'm itching to start a new project. (Yeh, it's going to be another car.)
Seeing as I've already spent so much time on this one I wanted to have it viewable in real-time 3d as well so I took the extra task of learning how to build it for Unity webplayer. Since this was my first time with Unity, I used one of the example projects for the most part but I did try to add some of my own elements in there though. Maybe I should get into programming a bit more... I got some ideas for my next project that I'd like to try producing.
Link to Unity webplayer: https://dl.dropboxusercontent.com/u/21585227/Public.html
It's 149MB.
I'd be keen to hear (read) any comments and criticism you might have.
Bonus pic (which I posted in the WAYWO some time at the end of 2014). 3D0+Photoshop.
Thought I'd start a thread now that I got this project done. I'm really bad at documenting my progress as I go but I'll try to include the most important milestones here. I'll kick off with a few 3DO renders.
The initial idea was to just model an interesting object and then call it done. I got inspired by a photo of a race car fitted with a radial engine so I went with that. The reference image is lower down on this post. This is the engine at 600k tris, ibl clay material and rendered in mental ray.
After that, I got the idea to try making a game model "hero asset" and try normal map baking. I hadn't done any of that before so this turned into a learning project as well. And then the whole project started expanding.
Since I was already going ahead with all the normal maps I thought I'd add some more detail using nDo2 which I had used a bit at work. I'm actually quite proud on how the engine front cover engraving turned out.
Next up was trying out the Quixel DDO trial on texturing the engine. Such a fun way of working and I definitely left me wanting to try it out on another project later on!
It took friggin' ages to the normal map bakes on the first parts I did. Engine ended up having 15k tris after the normal map bakes.
[IMG]https://fc04.deviantart.net/fs71/f/2014/171/8/3/radial_8___game_model_by_ollite20- d7n6zeo.jpg[/IMG]
Like I mentioned earlier the project started expanding and just having an engine didn't feel enough so I decided to create something around it and started gathering interesting reference photos. I liked the aesthetics of the 1930's grand prix racers with the exposed steering, suspension and wheels so I went with that as the main theme.
I had had the main inspiration photo for the engine on my hard drive for a long while before starting on modeling the engine. (The car is the 1935 Monaco Trossi which I later found out was horrible to drive because of the brutal front-heavy weight distribution.)
Everything is arranged onto one texture sheet and textured in Quixel Suite. Not fully satisfied with some of the materials in there but that'll do. The project file somehow got corrupted at some point so the only way to edit the materials would be to do them all over again. Textures' size is 4096x4096.
Most important ones. There's AO and opacity too but they're not that interesting to look at so I'll leave those out for now.
Just the mesh (3D0):
Mesh + normal map (3D0):
Mesh + normals + spec + gloss (3D0):
Everything (3D0):
With UVs (3D0):
It's been a long but fun project and I've learned a lot while working on this. Many firsts in this for me. The seat leather wrinkles and the cloth wrapped around the exhaust within the engine bay, among other things mentioned, are my first try at sculpting as well. Sure, there's still a few things I'm not 100% satisfied but I gotta move on. I'm itching to start a new project. (Yeh, it's going to be another car.)
Seeing as I've already spent so much time on this one I wanted to have it viewable in real-time 3d as well so I took the extra task of learning how to build it for Unity webplayer. Since this was my first time with Unity, I used one of the example projects for the most part but I did try to add some of my own elements in there though. Maybe I should get into programming a bit more... I got some ideas for my next project that I'd like to try producing.
Link to Unity webplayer: https://dl.dropboxusercontent.com/u/21585227/Public.html
It's 149MB.
I'd be keen to hear (read) any comments and criticism you might have.
Bonus pic (which I posted in the WAYWO some time at the end of 2014). 3D0+Photoshop.
ollitei
3 ·
Re: WIP Stylised Fantasy Character
Some progress on the hair and other bits and pieces. I need to clean up the shapes on the back of the head.
aniceto
3 ·