Hello Polycounters,
as the title suggests, this thread is about how to achieve that rly crips detail in lowres textures to avoid that blurry mess.
Therefore iam looking for crips and sharp brushes that seem to be hard to find.
To show you the problem i got with most ps bruhes:
![crips_blurry.jpg](http://www.blindest.de/crips_blurry.jpg)
The crips one is a brush that stays even sharp if you size the texture up by 200%
However the blurry one is how most PS-Brushes out there behave. Right now i work around it to create a similar effect but it simply takes too much time.
1st QUESTION:
How do you create a sharp Brush like that or where can you find those?
2nd QUESTION:
Is it necessarily better using mostly sharp brushes for textures or is this just important when working 1:1 with lowres textures to avoid losing too much pixel infomation.
Atm i only use the more blurry brushes to break up the crispness or cleaness since you cant get sharpness back if not starting sharp.
Feel free to share your texture workflow and fave brushes.
Any thoughts and input on that matter is welcome.
Replies
Yea probably the fastest way to make your own to fit your needs.
Might post some examples once i improved my workflow.
Are you from Benvista or you just like to advertise an amazing product?
So you think painting 1:1 isnt needed? The benefit in working 1:1 for me is that it makes me work more accurate. When working sized up i feel you tend to develop a sloppy style.
Can you show some examples how it looks sized down with that program compared to PS? I think i might need to test it to confirm. I guess you see it with 1k to 2k textures if its sized down in PS. Its just not feeling as crisp, but might try that program you mentioned.
@Noors I didnt actually know about the smart sharpen, dunno how i missed it, always used unsharp mask. Seems to work quite well if sharpen is needed.
This is zoomed in to an extreme to see what it actually does. I guess depends on your texture which algorithm you pick but i dislike having little control over the end result.
I actually would but iam not able to since i havent produced much outside of work lately. So obviously cant show any textures created due to copyright issues.
It actually helped a lot to know about these things mentioned here since i feel these were one of my particular problems since it slowed down my texturing process to keep it crisp and fitting to the rest of textures. As soon as i got something to show i ll just post it.
Yea the smart sharpen hint actually helped a lot since i wasnt aware of it. I always thought its the same as unsharpen but it seems to actually create the results i was looking for so iam happy to know about it now.
Will put everything new to use and post the results here.
No, I'm not for sure! May be my words sound like an AD, but I'm just a great fan of this program
Ok, I spent some time trying to compile a visual comparison of different approaches to imagery downsizing.
Here's a test image from free Ornament collection. Original resolution is 1950x1323.
I've chosen four areas of interest.
Left to right.
1) Linear visual elements + smooth color transition are.
2 and 3) Very fine details
4) Texture of stone with mid and fine details.
Then I've resized the image to 295x200 (don't mind the non-square size of the image - it just an example) using 7 different algorithms.
Now about the downsize approaches (top to bottom in the table).
1) PS bicubic - is a standard Photoshop solution since the very early releases. It's most popular and versatile.
Gives smooth details - good for photography but not textures.
2) Since CS3 (? - I'm not sure) Photoshop received a new algorithm to create sharper downsize of photos. The result is much better than simple bicubic resize.
3) Another popular method to raise a sharpness of downsized image is to apply Unsharp mask filter before resizing. I've tried different setting and 4 px @ 90% seemed to me pretty good. But the negative effect of this approach is the undesired contrast contours around sharp details. It wouldn't hard a photo but texture gets odd visible outlines.
Next four algorithms (most useful IMHO) are from Photozoom Pro 4. There are several others, but not that nice.
4) Standard S-Spline Max Generic resizer gives even more blurry result than PS Bicubic. Not good.
5) S-Spline Max Photo Extra Detailed gives a bit sharper result than Generic (almost similar).
6) S-Spline XL Photo Extra Detailed gives very crisp details. May be useful for some noisy textures but in most cases gives odd coarse effect.
7) My favorite - S-Spline Max Graphics Extra Detailed. Gives the most adequate result with almost perfect balance of sharpness in contours and texture. It handles the tiniest details and lines in the image. Also it doesn't increase the contrast on the sharp edges.
You may agree or not that the last algorithm gives the best result of all but personally I find it most cool!
Anyway here you can see how different you downsized texture may look. And it feels sorry for getting your work look blurry and loose it's fine details.
Hope my little review will help
P.S. That's why I think that 1:1 texture creation is an anachronism. ))
You made me curious so i ve to give it a shot.
Whats the biggest advantage for you to work double size and resize later?
It surely takes more space, updates slower in the 3d package or realtime. And as stated i tend to work sloppier if i feel it gets sized down anyway.
I guess at the end of the day it doesnt rly matter if you work 1:1 or 2:1 as long as you feel comfy and the end result is nice.
I can't help agreeing
The assets will be used in marketing, artbooks, and other non-game productions.
The assets will be used in cutscenes.
The assets will be used on the PC where you want to be able to tighten up the graphics.
The memory budgets we have for textures may change, and all of the sudden I have an extra meg available. Or maybe I lose a meg. Either way, having higher resolution assets with which to work while I'm refactoring my stuff gives me more flexibility.
Working 1:1 gives you a better end result - I believe this strongly. The problem is that it is inflexible, and game development is anything but rigid.
Thanks for the scaling comparison, I agree that #7 looks the best.
Yea that makes perfect sense. I guess it also depends on the studio and kinda game you work on. Different workflows, different usage of assets, FP/TP/Racing/RTS, platform etc.
Every man to his own taste.
Word!