You upload a high-quality photo and hit convert. The 3D preview looks promising. You download the STL, slice it, and hit print. But when the print finishes, the result looks nothing like what you expected. The details are mushy, the edges are stepped, and that crisp logo you started with is barely recognisable.
If this has happened to you, the problem is almost certainly not with your printer or your slicer. It is with your source image. The quality of your final 3D print is determined long before it reaches the build plate — it is baked into the resolution, contrast, and format of the image you start with.
Let me explain exactly how image characteristics affect STL quality, and more importantly, how to choose and prepare images that will print beautifully every time.
How Heightmap Conversion Works at the Pixel Level
To understand why resolution matters, you need a basic picture of what happens when a tool like ImageToSTL converts your image to a 3D model.
The converter reads your image pixel by pixel. For each pixel, it calculates the brightness — a value between 0 (pure black) and 255 (pure white). That brightness value becomes a height. A bright pixel becomes a tall vertex in the 3D mesh. A dark pixel becomes a low one. The grid of all these vertices is then connected with triangular faces to form a continuous surface.
This means every pixel in your source image corresponds to a vertex in the 3D mesh. The more pixels you have, the more vertices the mesh contains, and the more detail it can represent. This is the direct relationship between image resolution and STL quality.
Key takeaway: Each pixel becomes one data point in your 3D model. More pixels = more detail. Fewer pixels = less detail. It is that straightforward.
The Practical Resolution Thresholds
Based on hundreds of conversions and prints, here are the practical resolution thresholds I have found:
| Image Size (pixels) | STL Quality | Best For |
|---|---|---|
| Under 200 × 200 | Poor | Simple geometric shapes only (circles, squares) |
| 500 × 500 | Good | Logos, text, bold icons, small keychains |
| 1000 × 1000 | Very Good | Portraits, detailed reliefs, medium-sized prints |
| 2000 × 2000+ | Excellent | Large lithophanes, fine art reproductions, professional work |
Note that the converter at ImageToSTL scales images down to a maximum of 500px on the longest side for performance reasons. This is more than enough for most keychains, badges, and small to medium prints. The heightmap does not need to match your printer's native resolution — the mesh resolution and the print resolution are two different things.
PNG vs JPG: Does the Format Really Matter?
Short answer: yes. Long answer: it depends on what you are converting.
PNG uses lossless compression. Every pixel is stored exactly as it was created. When the converter reads a PNG, it gets precise brightness values for every pixel. The heightmap is a faithful translation of your image with no compression artefacts.
JPG uses lossy compression. It throws away fine variations in colour and brightness to reduce file size. For photographs viewed on a screen, the difference is often invisible to the human eye. But the converter's pixel-reading algorithm is ruthlessly honest — it sees every compression artefact and turns it into surface texture. A JPG that looks fine on your phone screen can produce a 3D model that looks noisy and blotchy.
Rule of thumb: Use PNG whenever possible, especially for logos, text, and line art. For photographs, a high-quality JPG at minimal compression (settings of 10–12 in Photoshop, or "maximum" quality) is acceptable, but PNG is still better.
Contrast Is More Important Than Resolution
Here is something that surprises a lot of people: a 300 × 300 pixel image with strong contrast will produce a better 3D print than a 3000 × 3000 pixel image with flat, muddy contrast.
The heightmap converter translates brightness differences into physical depth. If your image has a wide range of brightness — pure whites next to pure blacks — the resulting model will have pronounced, clearly defined features. If your image is mostly mid-tones with subtle variations, the model will be flat and hard to read, regardless of how many pixels it has.
Think of it like this: resolution determines how many data points you have. Contrast determines how meaningful those data points are. High resolution with low contrast is like having a very detailed map of a flat plain — lots of information, nothing interesting to see. Moderate resolution with high contrast is like a simple topo map of a mountain range — fewer lines, but every one means something.
For the best results, aim for a histogram that uses the full brightness range. Your image should have true blacks, true whites, and smooth transitions between them. If your image's histogram is clustered in the middle, boost the contrast in any photo editor before converting.
The Max Height and Detail Trade-Off
The Max Height setting controls the vertical exaggeration of your model. A high Max Height (say, 10mm) makes every brightness difference more pronounced. A low Max Height (2mm) compresses everything into subtle relief.
Here is the trade-off: Max Height amplifies both signal and noise. If your image has good contrast and clean pixel data, a higher Max Height produces dramatic, impressive depth. If your image has compression artefacts or noise, a higher Max Height will make those artefacts more visible in the print.
Practical guidance per project type:
- Lithophanes: 2–3mm max height. You want subtle, smooth transitions for light to pass through.
- Keychains and badges: 3–4mm max height. Enough depth to be visible, thin enough to carry.
- Wall plaques and signs: 5–8mm max height. Good readable depth for text and logos.
- Coasters: 4–6mm max height. Deep enough to feel the design under a drink, thin enough to be practical.
- Dramatic display pieces: 8–15mm max height. Maximum depth for maximum visual impact.
How Image Noise Affects Surface Quality
Image noise is random variation in pixel brightness that does not correspond to real detail in your subject. It comes from camera sensors in low light, aggressive JPG compression, or scanning textured paper. In a 2D image, noise looks like grain. In a 3D model, noise looks like surface roughness — tiny random peaks and valleys that make the print look fuzzy.
The converter's smoothing setting is designed to reduce this. Smoothing averages each pixel's height with its neighbours, effectively applying a blur to the heightmap. This softens the transitions and reduces the visual impact of noise.
For low-noise images (clean PNGs, vector-derived images, well-lit photographs), use minimal smoothing to preserve sharp detail. For noisy images (high-ISO photos, scanned documents, highly compressed JPGs), use higher smoothing (3–5) to salvage a clean result.
But the best approach is to start with a clean image. Here is a quick pre-processing workflow:
- Denoise the image. Most photo editors have a "denoise" or "reduce noise" filter. Apply it gently — too much removes real detail.
- Apply a slight Gaussian blur. A blur radius of 0.5–1 pixel smooths out random pixel variations without destroying edges.
- Re-sharpen. After denoising and blurring, apply a light sharpening pass (unsharp mask, radius 0.5, amount 50%) to restore edge definition.
- Check the histogram. Make sure the brightness range spans from near-black to near-white.
Real-World Example: Same Image, Different Resolutions
Let me give you a concrete example from my own experience. I converted a company logo at three different resolutions and printed the results to compare.
The 150 × 150 version: The letters were recognisable but the edges were visibly stepped. Fine details in the icon were lost entirely. The print looked like a low-resolution bitmap blown up — blocky and unrefined.
The 500 × 500 version: Clean, sharp edges. All the icon details were present. The print looked professional. This is the sweet spot for most logo-based projects.
The 2000 × 2000 version: No visible improvement over the 500 × 500 version because the converter scaled it down internally. The extra pixels did not hurt, but they did not help either — they were simply discarded during processing.
The lesson: do not obsess over massive source images. A clean, well-contrasted image at 500–1000px on the longest side will produce excellent results for almost every project. What matters far more is the quality of those pixels — clean edges, good contrast, and no compression artefacts.
Image Preparation Checklist
Before you upload any image to a converter, run through this checklist:
- Resolution: At least 500px on the longest side. 1000px+ for detailed work.
- Format: PNG preferred. High-quality JPG acceptable for photos.
- Contrast: Histogram spans the full brightness range. True blacks and whites present.
- Noise: Minimal visible grain. Denoised if necessary.
- Background: Clean and uniform. Removed entirely for best results.
- Cropping: Subject fills most of the frame. Minimal empty space.
- Sharpness: Edges are crisp. No motion blur or soft focus on important details.
Frequently Asked Questions
Does a higher resolution image always produce a better STL?
Not always. Beyond a certain point — roughly 500–1000px on the longest side — the converter's internal downscaling means extra pixels are discarded. A lower-resolution image with excellent contrast and clean edges will outperform a high-resolution image with poor contrast and compression artefacts every time.
What is the best image format for STL conversion?
PNG is the best all-around choice. It is lossless, supports transparency, and preserves exact pixel values. For photographs, a high-quality JPG (compression setting of 10+ or "maximum") is acceptable. Avoid GIF (limited colours), BMP (large files), and heavily compressed JPGs.
Can I use a vector file (SVG, AI, EPS) to create an STL?
Most image-to-STL converters work with raster images (PNG, JPG). To use a vector file, export it as a high-resolution PNG first. Export at 1000 × 1000px or higher for a clean result. Vectors are actually ideal sources because they have perfectly sharp edges and no noise.
Why does my STL have jagged edges when the original image is smooth?
Jagged edges in the 3D model usually come from insufficient image resolution or insufficient smoothing. If your source image is a low-resolution JPG, each pixel is large and the transitions between pixels are abrupt. Try using a higher resolution image or increasing the smoothing setting in the converter.
How does the converter handle transparency in PNG files?
When "Remove Background" is enabled, the converter checks the alpha channel of your PNG. Pixels with an alpha value below 50% are treated as having zero height. This allows transparent backgrounds to become flat areas in the 3D model. For best results, export your PNG with a clean, fully transparent background.
Does colour information affect the STL output?
No. STL files contain geometry only — no colour, texture, or material data. The converter converts your image to grayscale internally and uses only the brightness values. Colour differences that have the same brightness (red and green at the same luminance, for example) will produce the same height.
Should I use black and white images or grayscale?
For logos, text, and line art, a pure black-and-white (binary) image gives the cleanest result. For photographs and continuous-tone images, use full grayscale with smooth transitions between shades. The converter handles both equally well — it reads the brightness of each pixel regardless of whether the image is binary or continuous.
The Bottom Line
Image resolution and STL quality are directly connected, but the relationship is simpler than most people think. You do not need massive, multi-megapixel images. You do need images that are clean, well-contrasted, and free of compression artefacts.
Think of the converter as a very literal translator. It takes what it sees and turns it into plastic. Give it a clear, high-quality image, and it will produce a clear, high-quality STL. Give it a noisy, compressed, low-contrast image, and no amount of slicer tweaking will fix the result.
A little effort at the image preparation stage — cropping, denoising, boosting contrast, choosing PNG over JPG — pays off enormously in the quality of your final print. And the best part is that once you develop the habit of preparing your images well, every conversion becomes more predictable and more satisfying.
Put it to the test
Upload a high-resolution PNG and see the difference a clean source image makes. Free, private, no sign-up required.
Open the Converter →