PNG Layer Requirements for NFT Art: Dimensions, Transparency & Naming
More generative collections get derailed by messy source files than by anything else. Wrong dimensions, missing transparency, inconsistent naming — none of it is glamorous, but get this part wrong and everything downstream breaks. Here's how to set up your layers so the generator (and you) never has to fight them.
01 — Why File Prep Is the Unsung Hero
Nobody dreams of becoming an NFT creator so they can obsess over pixel dimensions and file naming conventions. But a generative collection is really a stack of images being composited automatically, thousands of times over — and automation has zero tolerance for inconsistency. A single misaligned layer at position 1 will be misaligned at position 8,742 too.
Key point: Every problem in this guide is invisible if you only look at one or two traits in isolation. They only become obvious once you generate hundreds or thousands of combinations — which is exactly why it's worth getting right before you start, not after.
02 — Dimensions: Pick One Size and Never Deviate
Every trait across every layer needs to share the exact same pixel dimensions — the same width and the same height, down to the pixel. If your background is 2000×2000px and your hat trait is 1998×2002px, the generator will still stack them, but the misalignment will be visible the moment you zoom in.
Sensible size choices:
- 1000×1000px — smaller file sizes, faster generation, fine for most PFP collections
- 2000×2000px — the current sweet spot for most 2026 collections; sharp on modern displays without bloating file size too much
- 3000×3000px+ — only worth it for premium/1-of-1 style drops where print-quality resolution matters
03 — Transparency: The Invisible Glue Holding Everything Together
Every trait file needs a genuine alpha channel — the part of a PNG that defines which pixels are see-through. Without it, each layer would sit on top of the previous one as an opaque rectangle, hiding everything beneath it.
Common mistake: Exporting a PNG from certain design tools with a white or checkered background baked in, rather than true transparency. Always zoom in and check for the checkerboard pattern your editor uses to represent transparency before exporting — a flat white background that "looks transparent" in a thumbnail preview is the most common silent killer of a generation run.
04 — File Naming: Boring but Non-Negotiable
Your file names often become your trait names in the final metadata, so consistency here saves you an entirely separate headache later on marketplaces.
| Do | Don't |
|---|---|
| Gold_Chain.png | gold chain final v2.png |
| Laser_Eyes.png | Laser eyes (1).png |
| Blue.png | blue background.PNG |
The naming rules that actually matter:
- Pick one capitalisation style and stick to it across every single file
- No spaces at the start or end of a filename
- No version markers like "final," "v2," or "USE THIS ONE" left in the name
- Keep the file extension lowercase and consistent (.png, not .PNG)
Let the Generator Catch Prep Mistakes for You
Upload your layers and TheMintLab flags dimension mismatches and naming inconsistencies before you generate — free, in your browser.
Try the Generator →05 — Folder Structure: Set It Up Once, Thank Yourself Later
Organise your traits into one folder per layer — /Background/, /Body/, /Eyes/, /Hat/ — with only the trait files for that layer inside. Most generators, including TheMintLab, read folder names directly as layer names, so a clean structure means less manual configuration once you upload.
Frequently Asked Questions
What image format should NFT trait layers be in?
PNG with a true alpha channel (transparency). This is essential for layering traits cleanly on top of one another without opaque backgrounds hiding the layers underneath.
Do all my trait files need to be the exact same size?
Yes — every file across every layer needs identical pixel dimensions. Even a few pixels of difference can cause visible misalignment once layers are stacked.
Does file naming actually affect my final metadata?
In most generators, yes — file names are commonly used as the trait values in your metadata's attributes array. Messy file names can mean messy, unprofessional-looking trait labels on marketplaces.
What's the best resolution for a PFP-style NFT collection?
2000×2000px is a common standard in 2026 — sharp on modern screens without unnecessarily large file sizes. 1000×1000px is also widely used and generates faster for very large collections.
Boring Prep, Beautiful Results
None of this is exciting work, but it's the difference between a smooth generation run and hours spent hunting down why token #6,204 has a floating hat. Get your dimensions, transparency, and naming locked down before you touch a generator, and everything after this point gets dramatically easier.
TheMintLab's generator flags common file prep issues automatically on upload — free, no account required.