Takes low-resolution customer artwork — 72 DPI JPEGs, screenshots, web images — and upscales to print-ready 300 or 600 DPI without pixelation. Tuned for DTF transfers: preserves edge crispness which matters more than photographic detail at print size.
Try the upscaler →DTF printing requires 300 DPI minimum (600 DPI for fine detail like small text or photographic transfers). Customer-supplied artwork is almost always lower resolution — phone screenshots are 72 DPI, downloaded web images are 72-96 DPI, social media exports are 72 DPI. Printing these directly produces fuzzy, pixelated transfers.
The AI upscaler uses a neural network trained on DTF-specific artwork pairs (logos, illustrations, transfer designs at low and high resolution) to predict what the high-DPI version should look like. Unlike Photoshop's bicubic enlargement (which just stretches existing pixels), the AI model invents detail that didn't exist in the original — and gets it right because it's trained on millions of similar examples.
Generic AI upscalers (Topaz, Waifu2x, ESRGAN) are trained on photographs. They optimize for preserving photographic texture — skin, hair, scenery details. That's the wrong target for DTF.
DTF transfers care about edge crispness — sharp lines on logos, clean letterforms in text, hard edges on illustrations. A photographic upscaler will produce soft, washed-out edges that look blurry on the printed transfer.
DTFGSA's upscaler is trained specifically on DTF artwork. Edges stay sharp. Letterforms stay legible at small print sizes. Logos stay crisp. The model deliberately downweights photographic detail in favor of edge preservation — which is what matters for transfers.
| Input resolution | Target output | Quality |
|---|---|---|
| 72 DPI (web/phone) | 300 DPI | Excellent (~4x) |
| 96 DPI | 300 DPI | Excellent (~3x) |
| 150 DPI | 300 DPI | Excellent (2x) |
| 72 DPI | 600 DPI | Good (~8x — best for line art, logos) |
| 36 DPI (very low) | 300 DPI | Marginal — works for simple shapes |
| Under 200×200 pixels | any | Difficult — may need original artwork |
Pair with background removal: common workflow for memorial/pet portrait transfers — customer uploads a low-res phone photo with a busy background. AI background removal extracts the subject. AI upscaler bumps resolution to 600 DPI. Result: print-ready transparent PNG from a phone selfie in under 15 seconds total.
The AI upscaler is included in:
Free and pay-as-you-go don't include AI image tools. See full pricing comparison.
Yes. The AI upscaler takes 72 DPI input (typical phone/web image) and upscales to 300 or 600 DPI suitable for DTF printing. Tuned specifically for DTF artwork — preserves edge crispness rather than photographic detail. Available in Starter tier and above.
3-8 seconds per image depending on size. Small images (logos, badges) typically 3 seconds. Larger images (full-back designs, photos) up to 8 seconds. Result drops back into the gang sheet immediately, ready for nesting.
Best results up to 4x (72 DPI to 300 DPI). 8x (72 DPI to 600 DPI) works for line art and logos but may show softness on photographic content. 12x+ is generally too aggressive for any input — request the original high-res artwork from the customer if you need that range.
No. The white channel is generated from the upscaled artwork's alpha mask, with adaptive choke calculated based on the upscaled edge characteristics. You always get a correct white channel regardless of input resolution.
Upload a low-resolution image. Watch the AI bump it to 300 DPI in seconds. See the result in your gang sheet. Free to try.
Open the builder →