TL;DR: Start with the highest-resolution source you have. Use AI upscaling to hit TikTok's 720×1280 minimum, or push to 1080×1920 for future-proofing. Compress before export. Never upscale a heavily compressed or tiny original.
Low-res images tank TikTok ad performance. Meeting the platform minimum is just the starting line. Sharp, clear visuals are what actually hold attention in a fast-moving feed.
---
Why Upscaling Matters for TikTok Ads
Getting your resolution right is the first quality gate every TikTok ad must pass.
TikTok's recommended image resolution and platform minimum specs
Per TikTok For Business guidance, In-Feed Ads require a minimum of 720×1280 pixels. Global App Bundle ads share the same minimum. Accepted file formats are .png, .jpeg, and .jpg, with a file size cap of 100 MB. For extra headroom, target 1080×1920. That matches TikTok's recommended Full HD vertical resolution and renders cleanly on every screen size.
How low-resolution images hurt ad relevance scores and CTR
Crisp visuals improve ad relevance scores on paid platforms. Better relevance means better CTR and typically lower cost-per-click. A blurry image signals low quality before a user reads a single word.
When upscaling is the right solution vs. when to reshoot
Upscaling shines when you're adapting existing assets to new specs or repurposing content across channels. It cannot recover detail that was never captured. If your source file is smaller than roughly 300×300 pixels or riddled with compression artifacts, the right call is a reshoot.
---
Traditional Resizing vs. AI-Powered Upscaling
The method you choose determines whether your upscaled image looks sharp or smeared.
What traditional resizing does (and why it causes pixelation)
Traditional resizing spreads existing pixels across more space. No new information is added. The result is blur, pixelation, and soft edges that look unprofessional at full screen.
How AI upscaling uses machine learning to preserve detail
AI upscaling analyzes patterns, edges, and textures in the original image. It then reconstructs missing detail at the higher resolution. The output stays sharp because the model is predicting and rebuilding structure, not just stretching pixels.
Key metrics: quality retention, artifact prevention, file size balance
Good AI upscaling preserves fine edges without over-sharpening. It avoids the "plastic" look common with aggressive processing. After upscaling, compress the output to keep load times fast without sacrificing visible quality.
---
The Best Approach: AI Upscaling Step by Step
Follow these four steps and you'll hit TikTok's specs without degrading your creative.
Step 1: Assess your source image quality and starting resolution
Check the pixel dimensions before you start. Anything below 360×640 is a risky source. Look for compression artifacts like blocky areas or color banding. The cleaner and larger your source, the better your upscaled result.
Step 2: Choose your target resolution
For TikTok In-Feed Ads, 720×1280 is the hard minimum. Targeting 1080×1920 gives you flexibility across placements. A 2x to 4x upscale from a mid-resolution source typically hits the best balance of detail and file size.
Step 3: Use an AI upscaler with platform awareness
Choose a tool that rebuilds detail without adding unnatural edges or over-sharpening artifacts. Coinis Revise includes AI Upscale as one of its seven editing capabilities. Upload your image, apply AI Upscale, and get a higher-resolution output ready for TikTok's specs. Note. Coinis currently publishes directly to Meta (Facebook and Instagram). TikTok direct publishing is on the roadmap. Revise handles the creative work for any platform, including TikTok.
Step 4: Compress and export for fast loading without quality loss
After upscaling, compress to the smallest file size that holds visual quality. Keep the file under 100 MB for TikTok. Export as .png when you need transparency or .jpg for smaller file sizes on photographic content.
---
TikTok Image Requirements at a Glance
Know the specs cold before you launch anything.
In-Feed Ads: 720×1280 pixels (or 1080×1920 for future-proofing)
TikTok For Business recommends shooting vertically and keeping key content in the center safe zone. TikTok's description text, profile logo, and action buttons overlay the bottom and edges of the frame.
Global App Bundle: 720×1280 pixels, under 100 MB
Same resolution as In-Feed. File size is capped at 100 MB. Formats are .png, .jpeg, or .jpg only.
File formats and safe zones for text and product placement
Keep all critical text and product imagery in the central vertical strip. Per TikTok's business guidance, TikTok's native UI elements crop and overlay the edges. Anything placed near the frame edges risks being hidden.
---
Avoid Common Upscaling Pitfalls
A few mistakes show up again and again. Skip them and save time.
Never upscale extremely small or heavily compressed source images
AI upscaling is powerful. It cannot invent detail that never existed. A 200×200 pixel product shot with heavy JPEG compression will still look rough at 1080×1920, even after the best AI pass.
Watch for over-sharpening artifacts and unnatural edges
Some upscalers apply too much sharpening. The result looks artificial and crunchy. Check hair, fabric textures, and curved edges in your preview. If they look harsh or outlined, the processing is too aggressive.
Test upscaled images on mobile and desktop before launching
TikTok is mobile-first. Preview your upscaled image on a phone screen before going live. What looks clean on a desktop monitor can reveal artifacts on a 6-inch display.
---
Or skip the steps.
Coinis Revise edits any ad image with AI. Move text. Change text. Swap colors. Erase objects. Translate to any language. One click each.
No design skills. No Photoshop. One click.
15 AI tokens a month. No credit card.
Frequently Asked Questions
What is the minimum image resolution for TikTok In-Feed Ads?
TikTok In-Feed Ads require a minimum resolution of 720×1280 pixels. For future-proofing across placements, targeting 1080×1920 is a better choice. Accepted file formats are .png, .jpeg, and .jpg, with a 100 MB file size cap.
Can AI upscaling fix a heavily compressed or very small image?
AI upscaling significantly improves mid-resolution images by reconstructing detail using machine learning. But it cannot add detail that was never captured in the original. Very small files (under roughly 300×300 pixels) or images with heavy compression artifacts will still show quality issues after upscaling. Start with the best source you have.