How much can AI upscale increase image size before quality drops?

AI upscaling is excellent at 2-4× the original size. At 4-8× the results remain usable for most purposes but artifacts may become visible up close. Past 8×, even the best AI models invent details that look unnatural at print size. Plan for the 4× ceiling for premium quality; 8× for distance-viewing prints.

Can I just change the DPI to make my image bigger?

No. DPI metadata is a tag that tells printers how many pixels per inch to use; it doesn't add pixels. A 1,000-pixel image stamped at 300 DPI prints at 3.3 inches sharp; the same image stamped at 72 DPI prints at 14 inches but at very low quality. To genuinely make an image bigger for print, you need more actual pixels (AI upscaling).

What's the cheapest way to AI upscale an image?

Several affordable AI upscaling options exist. Free tools like waifu2x or Upscayl are open-source and run locally, so they incur no costs and protect sensitive artwork. Paid platforms like Replicate charge about $0.002 per image, which can add up but offer consistent quality and speed. Services such as Ratio Ready integrate AI upscaling into print workflows, ideal for frequent Etsy sellers. For instance, a 500x500 pixel image for a 10x10 inch print can be upscaled with waifu2x to meet the 300 DPI standard needed for sharp printing. Occasional users will find free tools sufficient, while frequent printers benefit from integrated paid tools that streamline processes and boost efficiency.

Should I upscale before or after cropping?

Always crop your image before upscaling. Cropping reduces the pixel area you enlarge, which preserves quality and processing resources. For example, starting with a 1000x1000 pixel image and cropping to the desired 5x5 inch print section might leave 600x600 pixels. Upscaling this smaller, precise section to 300 DPI ensures better quality and less wasted effort. Upscaling first risks enlarging unwanted parts, increasing file size and reducing control over composition. Crop to exact print dimensions, then upscale to reach the proper pixel density. This method ensures your print matches desired specs with minimal quality loss or excess file size.

Will AI upscaling change the look of my image?

Modern AI upscalers preserve composition, color, and overall appearance — but they may slightly modify fine details (especially text, faces, and very thin lines). For typical wall art content (photographs, illustrations, abstracts), the differences are imperceptible at viewing distance. For images where every pixel matters (technical drawings, text-heavy posters), test the upscale output before committing.

Can I upscale a 1024-pixel Midjourney image to 24x36 print?

Yes, but it's a stretch. 1024 pixels needs 7-10× upscaling to reach 24x36 at 300 DPI. AI upscalers can do this but at the upper end of what produces clean results. Better path: generate at 2,048+ pixels in Midjourney (most plans support this) so the required upscale is only 4×.

What if I can't upscale and can't reduce the print size?

If you can’t upscale your image and reducing print size isn’t an option, consider printing at a lower DPI, like 200 DPI instead of the ideal 300 DPI. This compromise maintains decent quality for larger prints viewed at a distance, such as posters or wall art, although some detail loss occurs. If the image resolution is too low to produce acceptable quality even at reduced DPI, it’s better to offer a refund rather than ship a poor-quality print. Avoiding low-quality products protects your store’s reputation and fosters customer trust, which is vital for your long-term business success.

How do I know my upscale was successful?

Compare the upscaled file's pixel count against your target. If you needed 7,200 × 10,800 and got 7,200 × 10,800 (or larger), the upscale hit the target. Check sharpness at 100% zoom in Photoshop / Preview — at 100%, the image should look crisp and detailed, not mushy or pixelated. If it looks soft at 100%, the upscaler didn't add real detail.

Troubleshooting

Your image is too small for printing — here's what to do

An image too small to print typically requires resizing to at least 300 DPI or dimensions such as 3300x4200 pixels for quality output on platforms like Etsy and Printify. To fix a low res image, use methods like vector tracing, AI upscaling, or sourcing higher resolution files to avoid pixelation and maintain print clarity.

6 min read April 26, 2026

"Image too small for printing" is the most common print-prep problem for Etsy sellers, photographers, and digital download buyers. The fix depends on three things: how much smaller your image is, what the print is for, and how much quality you can compromise. There are five practical paths — AI upscaling, target size reduction, DPI compromise, source replacement, or order decline.

Most files fall short by 2-4× on the long edge. AI upscaling handles this range cleanly. Files that fall short by 8× or more usually need a different source image entirely — even AI can't invent that much detail without artifacts.

First: how short is your image, exactly?

Before fixing, measure the gap. Three pieces of information give you the full picture:

What you have: the source image's pixel dimensions (e.g., 2,000 × 1,500 = 3 MP).
What you need: the target print size in pixels at 300 DPI (e.g., 24x36 inches at 300 DPI = 7,200 × 10,800).
The shortfall ratio: divide what you need by what you have on the long edge. (10,800 / 2,000 = 5.4× short).

A 1-2× shortfall is easily fixable with AI upscaling and you'll get sharp results. A 3-4× shortfall is the upper end of clean AI upscaling. A 5-8× shortfall is doable but with visible artifacts. Past 8× you should use a different source image. Use the megapixel calculator to compute your shortfall in seconds.

The five fixes (in order of common use)

AI upscale to fill the gap. The most common solution for shortfalls of 1-4×. AI upscalers add genuine new pixels with realistic detail (not blurry interpolation). Costs less than $0.05 per image and runs in 30-60 seconds.
Reduce the target print size. If you wanted 24x36 but your image only supports 16x20 sharp, list as 16x20 instead. Easy, free, and the print quality stays at 300 DPI — the customer just gets a smaller print.
Accept a lower DPI compromise. 200 DPI is acceptable for prints viewed from across a room (large posters, wall art above sofas). You can print 24x36 from a smaller source at 200 DPI — the eye can't easily distinguish 200 from 300 at typical viewing distance.
Replace the source. If the shortfall is huge (5×+), find a different source. Stock photo (paid or free), re-shoot at higher resolution, or use a different design. Better to ship the right file than to AI-upscale beyond what produces clean results.
Decline the order or refund. When none of the above work and the buyer's expectations can't be met. Better to refund a $5 order than ship a soft 24x36 and earn a one-star review that costs you future sales.

AI upscaling explained — what it does and doesn't do

AI upscaling uses machine learning models trained on millions of images to "guess" what high-resolution detail should look like in places where the source image only has low-resolution data. The good models (ESRGAN, alexgenovese, WaveSpeed) produce genuinely sharp output that's nearly indistinguishable from native high-resolution capture.

What AI upscaling does well:

2× and 4× upscales of photographic content (landscapes, portraits, still life)
Botanical illustrations, abstract art, geometric patterns
Single-subject images with clear focal points
Source images with minor compression artifacts

What AI upscaling struggles with:

Text content (especially small fonts — the AI may "invent" wrong characters)
Faces with critical detail (the AI may shift facial features in subtle, sometimes unflattering ways)
Source images with severe noise, blur, or compression artifacts
8×+ upscales (artifacts become visible)
Heavy line art with very thin strokes

For Etsy wall art — mostly photographic, illustrated, or design content — AI upscaling at 2-4× produces results indistinguishable from native captures at typical viewing distance. This is the right tool for most "image too small" situations.

The fix workflow at a glance

Three-step diagram showing how to fix a small image for printing: source 800x600 px, AI upscale 4x, output 3,200x2,400 px sharp — The standard fix — AI upscale takes a small source to print-ready dimensions in seconds.

Which fix applies to your specific situation

"My phone photo isn't big enough for a 24x36 print." AI upscale 2-4× (works well for photographic content). A 12 MP phone photo upscaled 4× becomes 48 MP — close to the 78 MP needed and visually indistinguishable from native at typical viewing distance.
"My Midjourney/AI-generated image is 1024 pixels." AI upscale 4× or 8×. AI-generated content typically responds well to AI upscaling because the source already comes from generative models. Plan for the 4× ceiling for clean results; 8× works for distance viewing.
"A buyer wants a custom commission size that needs more pixels than my source." Either AI upscale (if shortfall is ≤4×) or decline the size and offer a smaller alternative. Don't promise sizes you can't deliver sharp.
"My old scan or family photo is too small." AI upscale 2-4× usually works. Older photos benefit particularly because AI can fill in detail lost to scanning artifacts. 8×+ tends to invent details that don't match the original photo's character — not always desirable for personal photos.
"My stock photo is the wrong size for my print." Either AI upscale (if shortfall is small) or buy a higher-resolution version. Most stock photo sites offer multiple resolutions; check whether a larger version is available before upscaling.

Mistakes to avoid when fixing a too-small image

1. Just changing the DPI metadata to 300 without adding pixels

Stamping 300 DPI on a 1,000-pixel image makes it print at 3 inches sharp, not at 8 inches sharp. DPI metadata doesn't add pixels — it changes how the printer interprets the pixels you have. To genuinely fix a too-small image, you need more actual pixels (AI upscaling), not different metadata.

2. Using Photoshop's "Bicubic Smoother" instead of AI upscaling

Photoshop's interpolation produces visibly soft, mushy enlargements. AI upscalers produce genuinely sharp output. The cost difference is small ($0.01-0.05 per image vs free in Photoshop) but the quality difference is enormous. Always AI upscale, never bicubic.

3. Trying to AI upscale 10× or more

Even the best AI upscalers start to invent unnatural details past 8× upscaling. A 1 MP source upscaled 16× produces images that look impressive at thumbnail size but reveal AI artifacts at print size. Cap your upscale at 4-8× depending on content type.

4. Promising a print size you can't deliver sharp

If your source can only support 16x20 sharp, don't list 24x36 as an option. Buyers expect 300 DPI for the size you advertise; delivering soft prints generates refund requests and bad reviews. Either upscale to genuinely support the size or list at the size your source supports.

5. Skipping the diagnostic step

"Just AI upscale it" without first measuring the shortfall leads to over-upscaling (wasted compute, lower quality) or under-upscaling (still too small after the upscale). Always check the gap first using the megapixel calculator — takes 30 seconds, saves rework.

Frequently asked questions

Fix any too-small image automatically

Ratio Ready upscales, crops, and DPI-stamps in one upload. 50 Creative Credits free on signup.

Try AI Upscaler