What it makes
Every generator here makes an image from a prompt, starts from an image you upload, matches a reference style and edits what it made, so none of that earns a point. What is left is what separates them: how large the image comes out, whether text inside it is readable, keeping a character the same across images, steering with a pose or a sketch, training a style on your own pictures, extending past the frame, upscaling, cutting out a background and true vectors.
Highest resolution it outputs6 pts
✓Leonardo.Aiover 4k◐Adobe Firefly4k◐Stability AI4k◐Black Forest Labs2k◐Ideogram2k◐Midjourney2k
Renders readable text in an image4.8 pts
✓Black Forest Labs✓Ideogram✓Leonardo.Ai✓Midjourney✓Stability AI
Keeps a character the same across images4.5 pts
✓Adobe Firefly✓Black Forest Labs✓Ideogram✓Leonardo.Ai✓Midjourney✕Recraft✕Stability AI
Controlled by a pose, sketch or depth map4.2 pts
✓Adobe Firefly✓Black Forest Labs✓Leonardo.Ai✓Stability AI✕Ideogram✕Midjourney✕Recraft
Train a model or style on your own images3.9 pts
✓Adobe Firefly✓Black Forest Labs✓Ideogram✓Leonardo.Ai✓Midjourney✓Recraft✓Stability AI
Extends an image past its frame2.7 pts
✓Adobe Firefly✓Black Forest Labs✓Ideogram✓Midjourney✓Recraft✓Stability AI
AI upscaling1.8 pts
✓Adobe Firefly✓Ideogram✓Leonardo.Ai✓Midjourney✓Recraft✓Stability AI
One-action background removal1.2 pts
✓Adobe Firefly✓Ideogram✓Leonardo.Ai✓Recraft✓Stability AI✕Black Forest Labs✕Midjourney
Exports true vector files0.9 pts
✓Adobe Firefly✓Recraft✕Black Forest Labs✕Ideogram✕Leonardo.Ai✕Midjourney✕Stability AI
Highest here Leonardo.Ai 9.2 · lowest Recraft 5.3 out of 10