This is an AI video generation comparison for
image-to-video 
prompt:
A rugged, sunlit close-up of a man sitting outdoors, his bare arm turned toward the light. On his shoulder, a bold black tattoo of a roaring dinosaur is sharply defined. The man’s face comes into brief focus as he notices movement beneath the ink in his arm's tattoo - eyes narrowing in disbelief as he stares down at it. Focus shifts entirely to the tattoo. The dinosaur inside begins to move eyes curiously, its nostrils flaring as though catching a scent of something delicious. the tattoo ink...
Log in to see full prompt.
Tested: October 19, 2025
Main issue with this prompt generation is usually dino having a tiny pizza slice in its mouth even before it lands on the ground and never touching the proper large pizza.
Tested: October 19, 2025
Overall, this model does rather well with this prompt.
Tested: October 19, 2025
Excellent 2d to 3d, man's emotions and prompt following.
Tested: October 19, 2025
Almost perfect, except for how dinosaur holding the pizza slice.
Tested: October 19, 2025
Hailuo 02 does ok here. Man's disbelief emotion is missing though, but kind of an interesting touch him closing his eyes before dino is about to burst out of tattoo.
Tested: October 20, 2025
Two attempts had the same issue: inability to show how a tiny dino jumps out of the tattoo. It was able to poke out of it but it's almost like the model can't imagine it fitting into the arm only sticking its head through while actually being behind the arm.
Does the man’s facial expression visibly react to the tattoo moving (eye movement, disbelief)?
Does the tattoo animation show subtle internal motion before bursting (eye flicker, nostril flare, shimmer)?
When the tattoo bursts, do the black ink particles spray naturally and fade quickly without lingering artifacts?
Does the emerging T-rex appear to form smoothly from the dispersed ink (no abrupt pop-in or visual glitch)?
After the T-rex emerges, does the tattoo area remain clean except for the empty circular border?
Does the mini 3D T-rex land believably on the picnic blanket with proper shadow and weight?
Is the pizza slice visible and proportionate to the T-rex’s size?
Does the T-rex’s bite and carry motion look natural (jaw open/close sync, no clipping)?
Check out the results from GROK (Grok Imagine v0.9) vs Freepik (Veo 3.1) vs Wan (Online Platform) (Wan2.5 Preview) vs Kling AI (Kling 2.5 Turbo) vs Freepik (Hailuo 02) vs PixVerse (PixVerse V5) for similar or identical prompts side-by-side.
Split scene cat jumps from fryer to rock show
Holiday chaos bullet time effect
