Thanks for the replies

Last year i got me a new camera (Canon 400 digital rebel 10 Mpix) which is a big improovement from my old 4 Mpix camera. I also got a tripod and remote shutter as well as a small bunch of lenses which makes it possible to make good shots that i desperatelly need.
And to answer the questions, i often make use of real life photos to create all bases, though i also make lots of stuff in photoshop without any photo. If i use a photo as base than (lets say a foto from a concrete wall) than i paint stuff on top of that photo, but prior to that i do several things.
First i decide how the texture needs to look like, and what real life photos i need, for that reason i always look around on the street 24 hours a day to see if i see usefull stuff for future textures. Being on the lookout never stops, as said 24 hours a day lol.
When needed i go take the fotos to where i found a good object. Once home i transfer the foto to the computer. Than theres a fixed amount of steps i make:
- My camera takes foto in 3888 x 2592 but i need a square, which means i enlarge the canvas(Photoshop btw) to 3888 x 3888 and paint the missing parts using the clone tool.
- The next step is to make the texture tilable in all directions so it can repeat itself endlessly seamsless. For that i have a filter called HalfWrap that flipflops the layer i am working on. To create the actual tiling is also done using the clonetool.
- Unavoidably the photo gets damaged (read fuzzy) due to the cloning, and to solve that i use the Unsharp Mask filter to make it all sharp again. And due the fact that videocards make everything fuzzy again, you can make it emphesized sharp if needed, too little is sometimes worse than too sharp.
- If needed i also copy the result to an Alpha channel and use that to slightly bumpmap the texture. This will bring back more dept into the texture. But you gotta do that carefully, out of 100 steps bumpmapping i never go further than 3. So at best you gotta do a tiny bumpmap.
- The next step is to change the color, and for that i use Image---> Adjustment---> Selective color and have that set to Neutral while i toy around with the colorsliders. In very rare occasions this doesnt work and i use Image---> Adjustment---> Variations.
And that pretty much finishes the base. Any texture is made out of several bases and each base is made in a similar method as described above. As sidenote, each step is saved in a seperate layer, while i name the layer to what i have done. As example:
- 1st layer (bottom one): IMG_2123_A Tilable (IMG_2123 is name of the photo)
- 2nd layer: Unsharp 280, 1 (sharpening the photo)
- 3rd layer: Bumpmap 3
- 4th layer: Col -8, 0, +15, +25 (re-collor the foto)
- 5th layer: DecayedSConcrete04 which is also the name of the file itself (DecayedSConcrete04.PSD)
In that way i always can go back in history. For instance i could start off again from layer 3 and give it a different color so the file could be called DecayedSConcrete06.PSD
The actual textures is a collection of bases, and naming everything makes sure i always can track where it came from, and make variations later on from that base if needed.
Than i collect all bases needed for a texture in oné file (normally i start out with 2048 x 2048 pixels base) Than i am gonna create all shapes present in the texture which is actually the most time consuming part !!!
To create depth i either use the Bevel and Emboss fuction, but preferably i go paint the shadows and highlights with the use of Burn and Dodge tool. For the shadow casting on the layers below i use the shadow function to drop shadows (though this is limited but most times it will works just fine.)
This texture (UTbase1 package, texturename: mlbPipeWall7TES:

Is entirely made from this foto:

These scratches:

And this circuitboard i got from the net:

Which points out, you can make pretty much anything, out of anything

.
.
.