How to avoid using photos in texturing?
Texturing has always been my weakest point. I've been reading articles related to texturing and have often found them saying "try to avoid using photos when you're texturing" now.. All I've done up to this point had been texturing with photos and using clone stamps with healing brushes and overlay layers with grunge etc.
So my thought is, what exactly do they mean by this. Will I be more successful doing realistic painting of the texture I want to achieve? Is hand painting textures valuable to a realism oriented artist? (Also I have very little drawing skills and almost no coloring/painting skills) .
So my question is sort of spread through this . should I avoid using all these photos in my creations? Or are thru just saying copy/pasting a photo is bad. How valuable is learning to hand paint textures?
So my thought is, what exactly do they mean by this. Will I be more successful doing realistic painting of the texture I want to achieve? Is hand painting textures valuable to a realism oriented artist? (Also I have very little drawing skills and almost no coloring/painting skills) .
So my question is sort of spread through this . should I avoid using all these photos in my creations? Or are thru just saying copy/pasting a photo is bad. How valuable is learning to hand paint textures?
Replies
I actually don't really know where this "don't use photo's" is coming from.
It's just another tool in your arsenal to use, and a really good one if you use it well.
just my 2 cents.
Read this:
http://oesterkilde.dk/racer445.html
And have a look at polygoo's article in the vertex 2 magazine.
Or also see the segment where Jonathan Fletcher goes over the texturing off the huntress, there is also something about using photo's in there.
HOWEVER
I do this purely as a learning exercise, it is quicker, easier, and more often than not better to photosource if you're doing anything realistic.
Using a photo as a base is fine, as long as you can neutralize any real world flaws with it, like imperfect lighting. Most of my photo sourced textures are probably 90% adjustments and hand painting.
edit: length
I've use alot of photos when I started to studying texture
it's like . . .:
but at the end of the day, my objects look like it wrapped by some shiny, flashy paper. Definitely, some things is wrong here
Now, the only way that I know about making good textures is "understanding your textures". What I mean by that is you have to learn about what textures you'll create, but don't worry the studying process is very simple. Just find good References and staring at it, good references is all around you and on internet too.
If you staring long enough you'll see alot of things going on in a texture : color, lightning, shadows, specular, glossiness, surface details, . . .. and they're different on each object.
Every single small detail ( crack, painted, scratch, grunge,. . . .) has it own anatomy depend on how the object was made ?, how old is it ?, what is it functionality?, which environment is it in?,...
You won't get those valuable infomation just by blending some photos
I'm sure there are many great tutorials that will explain every things to you, just take a look, and good luck with your textures !
Also more often than not I find it is better to desaturate and level photorefs, and use them as a layer mask on hand painted detail to add extra texture to them.
And I can see the confusion, yes if you just grab some stuff from cgtextures, copy paste the photo straight in, and choose a blending mode, and that's the only thing you do, then yes, using photo's isn't gonna be all that great.
So just because the "lazy" method does not give great results it doesn't mean that using photo's is bad.
a small thing I made for someone a while ago:
http://img208.imageshack.us/img208/5046/98la.jpg
But to further illustrate it:
Let's say you make some sort of worn rubber
bad way:
-Find 2 photo's of rubber, copy paste them, overlay -> done
good way: (I mean I'm learning to, this isn't the only/right way ofcourse and is gonna depend a lot on the exact material)
-Find a photo with not to much contrast, just large variation, put this on at a really low opacity (like 2-3%) to build up a base, maybe here 1 or 2 combined, depending on what you are going for, this adds a bit of a base, but shouldn't be very noticable.
-Maybe find some colored noise photo, add that in subtle. (preferred over PS noises or other computer generated maps)
-Now you maybe want some parts where it has been rubbed a lot, so you find a texture with some nice large scuffs, you isolate those, delete the bad parts, make it b/w, and make a mask out of it for a solid fill layer, place them where you want them.
-After this, maybe some small scratches, find a photo with some small nice scratches, make a mask out of it (see picture), place them where you want, erase the parts you don't want or clonestamp the nice parts, etc.
-Rinse and repeat for the things you want.
But ofcourse it depends, if you want some dust in the crevices, you would start off with your AO, colorize it, make a mask, with a brush work that up a bit, or if you got a nice scratch brush, use that.
Because where do all those grunge/scratch brushes come from? photo's.
So unless you do all your handpainting with a standard round brush, you are using them in some way, and they give the same end result, just the way you get there is different
There are also examples of models where the handpainted parts are really obvious (in a negative way, where it just looks "painted/made"), but that also doesn't mean that painting scratches by hand is something you should never do.
So for me texturing is about, choosing the right values, right amount of wear, interesting shapes, good colour usage, and all that stuff.
And if we are talking about actual materials, being able to analyze those, being able to split it up into "layers" (for example after picking the correct base values, being able to recognize the characteristics of that material, what is unique to it, and translating that into your texture), and imo all of the above is what makes you good at texturing.
And because of that I find statements such as "avoid using photos in texturing" is correct in most of case" is kind of silly, they are one of the tools at your disposal, and just because it gets used incorrectly doesn't mean it isn't a good tool to have, let people figure it out by themselves what is the best way to achieve a certain thing.
I admit it's sound silly LOL, and because english is not my native language, I'm not sure if " correct " is strong word. I mean avoid using photos in texturing is a good advice for people who still confusing about how to use theme. ahhh, I remember to put (*) sign after my sentence, I just forgot to describe in the end of the post.
(*) When you can do a good textue without depending to photo resource, at that point you have a good understanding about texture, you'll know how to use them correctly
in fact, I use dozen of photos in my texture, and it has no certain rules, for example I've used "paper noise" for metal surface, . . ..
hope you get my idea
It's not whether or not use photos, is how and when to use them. Few examples:
1. All rocks/plaster/bricks.wood etc. are excellent candidates for photo source and generating maps out of them. Even most detailed sculpted normal map, will not give you the same level of detail as photo. Not because you are bad at sculpting, but because rendering (baking) process is not able to catch all details to texture.
Especially if you are after realistic look. Or more stylized-realistic look.
2. All highgly reflective surfaces are very bad candidates for using photos. You can use photos as masks in that case. I personally would recommend just using Substance Designer and screw all this hand-painting process all-together.
All I can say that for all metallic/reflective props with unique texture, I will never touch hand painting software again, I will just generate texture from scratch.
If there is some surface that by some astonishing accident is for example wood, I will photosource it and more likely create separate material for it, to achieve desire quality and texel density.
And most other props. Well I usually slap tillable texture on them anyway... I'm really to lazy to create unique textures for everything. Photo-sourced or not.
3. In the end, if it look, good, it look good. Noone really care how you achieved it (;.
Staying away from photos for <insert random stupid reason here> is silly.
Like perna says: why handicap yourself? Use whatever you have to use in order to achieve the best end-result.
But yea, you shouldn't just take a photo, toss an adjustment layer on it and think you are done - you need to understand basics (highlights/shadows, exposure, what makes a photo "flat", what all adjustment layers do etc.) Photos are just a tool as someone has already stated in this thread: you extract things from them, such as masks and other data from them (noise, grunge, scratches and other detail) and you pay close attention to the interaction between the diffuse and the specular. Think like a painter and make your textures in steps, starting with the big picture and moving down to the details as you go.