Ten hero props, every one of them beautifully baked, and not one surface in the folder that repeats. That is the environment art portfolio I get sent most weeks, and it is usually the reason the person who made it is still applying.
Nobody writes "no trim sheet" on a rejection. They write nothing. What they thought was: I cannot tell from this whether you can fill a level.
A prop and a corridor are two different jobs
A hero prop rewards unique UVs. One asset, one texture set, every pixel spent on that object and no other. That is the job most tutorials teach and the job most portfolios show, because it is the one where the pretty render comes out at the end.
A corridor punishes unique UVs. Forty metres of wall is not one asset, it is twenty modules that have to line up, share a material, and still not read as twenty copies of the same thing. Give each module its own texture set and you have twenty materials, twenty bakes, twenty things to fix when the art director changes the metal, and a set nobody can re-dress without going back through all of it. The trim sheet is the answer to that, and it has been the answer for fifteen years.
So when a reviewer opens a folder of hero props and finds nothing tiling, the missing trim sheet is not the problem in itself. It is the evidence. It says the applicant has never had to build the second wall.
The four-metre budget
We hold texel density at 10.24 pixels per centimetre across a set. That is 1,024 pixels to the metre, and it is the number that keeps a 4K set on a rifle looking like it belongs in the same world as a 2K set on a pot.
Run that number through a trim sheet and the design problem turns into arithmetic. A 4096 sheet at 10.24 px/cm is exactly four metres of surface. Not four metres of wall — four metres of everything the wall will ever need. Skirting, the panel run, the horizontal join, a pipe, a vent, a plain area, the capping detail at the top. Every one of those strips has to fit inside four metres, and every one has to run the full width of the sheet or it will not tile.
Which is why I lay a sheet out in centimetres before I open Designer:
- Decide the real-world height of every strip first. A 12.5 cm skirting, a 50 cm panel, a 1 m plain run.
- Convert at your density and round to a power of two. At 10.24 px/cm, 128 px is 12.5 cm, 256 px is 25 cm, 512 px is 50 cm, 1024 px is a metre. Boundaries that land on power-of-two rows let you snap UV shells to whole strips, and let the sheet mip down without one strip bleeding into its neighbour.
- Add the strips up. If they come to more than 4096 px, something comes off the sheet. Nothing gets squeezed.
- Put the strip you will use most in the middle of the sheet, not at the edge. You will be sliding shells up and down that axis for a week.
Four metres sounds generous until you spend 2048 px on one dramatic full-height panel. That is half the sheet, on the detail that appears twice in the whole level.
The plain strip is the one you use most
A trim sheet with detail on every strip looks superb flat, at 100 per cent, in the thumbnail. In a level it is noise: twenty metres of corridor where the eye has nowhere to rest and every module fights every other module for attention.
Give about a quarter of the sheet to a flat panel carrying nothing but roughness variation and a slight dirt gradient. That strip will end up covering more surface area than everything else on the sheet combined. A trim sheet is not a collection of your best details. It is a kit, and a kit needs empty space in it. If every strip is interesting, none of them are.
The corollary is the part juniors find harder: the detail you do include has to survive the distance it will be seen from. A 6 mm bolt head authored at 10.24 px/cm is six pixels. It is a smudge at two metres and it is gone at ten. That bolt belongs on a decal or on geometry, not on a sheet that has to serve a whole corridor.
Do not bake ambient occlusion into a trim sheet
I have argued before that the AO you bake for texturing is a tight contact shadow and not a global darkening. On a trim sheet the argument goes further: bake almost none of it into the surface.
A prop's AO is defensible because a prop has fixed neighbours. Its own geometry is the only thing occluding it, and the bake is true. A trim sheet has no fixed neighbours. The same 30 cm strip will sit at a floor join, run along a ceiling, appear upside down on a bulkhead and mirror around a doorway. Occlusion painted into it is a contact shadow drawn where, half the time, there is no contact — and it repeats at exactly the same intensity every 30 cm regardless of what the geometry underneath is doing. That is a large part of why modular sets read as wallpaper.
Let the engine do contact. Bake AO if you want it for masking, keep it out of the base colour, and leave the strips neutral enough to be reused in an orientation you have not thought of yet.
The second wall test
Here is the test I run on my own sheets, and it is the exact thing that makes a portfolio piece land.
Build the wall. Then build a second wall from the same sheet — different height, different opening, a corner instead of a flat run — and add no new textures. If you can, you built a system. If you cannot, you built a texture with stripes on it.
Show both. One wall proves you can texture. Two walls from one sheet prove you understand what the job is actually about, and once the sheet exists the second wall is an afternoon.
What goes in the folder
Four images and one line of text.
- The sheet flat, with the strips labelled in centimetres.
- The UV layout over the wireframe of the modules, so the reviewer can see where the shells sit.
- Three pieces built from it, one of them a corner or a doorway. The join is where trim sheets fail and it is where a lead looks first.
- A grey-lit shot of the assembled set with no post, so the shapes have to carry it on their own.
Then one line of text: sheet resolution, texel density, how many modules it covers, how many materials the finished set uses. That sentence does more for you than the beauty render, because it is the only part of the presentation written in the language the person hiring you thinks in.
One honest warning about the shortcut. I sell a Hardsurface Trimsheet Generator, a procedural Substance Painter tool that hands you something like sixty different hard-surface sheets. It earns its money on a deadline and it is the wrong thing to put in a junior portfolio, because the part being assessed is the layout decision and the generator makes that decision for you.
Where the opposite is still true
None of this applies to a still. In archviz and in film there is one camera, one distance and one render, and a trim sheet buys nothing while costing a day of planning. Unique UVs and a 4K set per surface are simply better there. It does not apply to a hero silhouette either. A sword with a chipped edge that the camera holds for four seconds should have its own bake, its own damage, its own everything.
And if the job you are applying for is props or characters, show props and characters. The gap I am describing is only a gap if the folder says environment art on the front of it.
