Flp
#1265 posted by
Kinn on 2016/03/05 16:02:44
that's a spicya meataballa
Just Wanted To Share A Little Progress.
#1266 posted by
flp on 2016/03/09 21:00:52
I'm currently texturing at a sluggish pace.
http://i.imgur.com/d3KFLaG.png
#1267 posted by
necros on 2016/03/09 21:58:38
Needs more brown I think?
Brown = Quake
#1268 posted by
flp on 2016/03/11 17:36:46
You are right. It's very hard to get the colors right. I don't even think there's very much blue in quake at all. I will change the weapon. I've tried color picking from the original textures to find base colors. Seems like quake in general is a world of brownish mud.
http://i.imgur.com/SpciqRk.png
Hmmmm
#1269 posted by
Kinn on 2016/03/11 17:57:53
Model looks great but I'm concerned that such polygon density will cause a lot of the detail to just collapse once you run it through the meat-grinder that is 8-bit MDL vertex precision...
Resolution Is Too High Imho
Oh Of Course
#1271 posted by
flp on 2016/03/11 18:18:58
These are just wip shots. The polycount is well in quakelimits and the model looks good in quake too.
The texture resolution is double the targetsize and also in truecolor because thats just the way I paint. There is a lot of detail that will get lost in translation. Luckily this includes some of the dullness and subpar painting.
Here's a screenshot of what it will look like after the conversion.
http://imgur.com/VHsPzeh
#1272 posted by
Kinn on 2016/03/11 18:23:11
The polycount is well in quakelimits
It's not about polycount - it's about vertex density, and you'll only see the collapsing verts once all the anims are done, not with a t-pose test.
However, you seem to know what you're doing so I'll stop being annoying now.
I Think I Understand.
#1273 posted by
flp on 2016/03/11 18:28:07
It's about vertices that are to close to each other?
I actually didn't think about this :( Well I'll just hope for the best then.
Looks Great!
#1274 posted by
mjb on 2016/03/11 18:31:43
Really liking the progress flp!
#1275 posted by
Kinn on 2016/03/11 18:35:58
Basically, imagine finding the maximum X, Y, and Z extents of the model, over all frames.
In each dimension, the vertex coordinates are then quantized to adopt one of 256 values fitting inside the maximum extents.
I don't know if I've explained it well, but if you have two vertices that are very close to each other relative to the model extents (remember it's over all frames, not per-frame), then they will collapse.
Got A Image
Quote: "I don't even think there's very much blue in quake at all."
http://www.simonoc.com/images/design/sp/1000cuts3l.jpg
Kinn
It should be possible to get rough aproximation with single frame by adding temp triangles on min/max values for each axis extent, correct?
Totes, Yeah
#1278 posted by
Kinn on 2016/03/11 19:55:45
If you can reasonably guess what the biggest mins/maxs are gonna be for the animated version.
Melee attacks and death anims are usually the worst offenders in my experience.
#1279 posted by
flp on 2016/03/11 19:58:35
@Kinn
thx for the clarification I didn't know that about quake. That's probably why these idle animations seem so wobbly most of the time. Cause the animations are so small that only a few vertices skip the treshold to the next interpolated point.
Well I'll make sure there are no van damme splits in my animations.
@PyroGXPilot
Are those stock textures. The enemies sure seem to pop out of the picture.
#1280 posted by
Joel B on 2016/03/11 20:07:46
Hah no they're not stock. :-)
There is tho an interesting amount of blue, red, and green in Quake... it just depends on the episode/theme (and they're all earthier tones of course).
For basemap style enemies "tan" is definitely the dominant environment.
#1282 posted by
Joel B on 2016/03/12 00:08:27
That's a great article BTW, it's cool that you take the time to write those up.
#1283 posted by
metlslime on 2016/03/12 00:38:51
yeah, i just read it for the first time too. Great, detailed post mortem.
#1284 posted by
sock on 2016/03/12 03:41:00
Wow, thank you @JohnnyLaw & @metlslime, always good to get feedback on my articles, took me forever to create it. Made my evening!
FLP, Re: Post 1271
would a touch less dithering look better on the "shirt" area?
#1286 posted by
flp on 2016/03/12 16:10:42
@Shamblernaut
The shirt area needs more detail. As of now the shirt is almost one solid color without dithering. I plan on some leather straps and more variety and then I can get rid of this awfulness. I tried a few different conversion algorithms but I might need to push pixels manually in the end.
I've also prematurely documented some of the process while procrastinating.
Here's an overview of the stages the modeling went through.
http://pnahratow.github.io/chainforcer-lowpoly-modeling-stages.html
Here is a timelapse video of some texture painting.
https://youtu.be/SlOZWpu9t7o
Great Work Flp
#1287 posted by
killpixel on 2016/03/12 17:09:17
Really digging everything you're doing.
Nice Article
Is there some sort of hybrid between nearest neighbour and dithering that works well?
It's been probably a decade since I had to use the (I'm showing my age a little here) reduce colour depth function in Paint Shop Pro.
Also, until I started watching your video just now, it never occurred to me just how silly it is shading on what really should be a dynamically lit model.
#1289 posted by
flp on 2016/03/12 20:42:29
@ Bloughsburgh, KillPixel
ty, very much!
@Shamblernaut
... it never occurred to me just how silly it is shading on what really should be a dynamically lit model
I must say I really disagree. Sure eventually you're going to be right. But given the years of ugly plasticy looking video game renderers we witnessed ever since normalmaps entered the jungle the handpainted artwork is one of the reasons why I like quake/old games so much.
Sadly, nowadays
this kind of stuff has drifted a bit into the cartoonish world of games like Dota or Lol.