This is an AI video generation comparison for
text-to-video 
prompt:
A delicate butterfly flutters into frame and lands on a purple flower, placed slightly off-center with a white middle, captured in extreme macro detail against a soft, greyish blurred backdrop. The fragile wings shimmer under faint sunlight as the camera holds steady. Then a rack focus sharpens the background - heavy tank barrels emerge, dust rising, soldiers’ legs rushing past as shouts echo through the street. War-torn street engulfed in rubble, smoke, and fire. A stark, cinematic portrait of...
Log in to see full prompt.
Tested: October 4, 2025
That's an amazing generation, even though rack focus isn't obvious, pull back happening straight up. For some reason, Veo likes to shrink the video's height for this prompt. Happened over and over.
Tested: October 4, 2025
Great cinematic result, rack focus implemented seamlessly
Tested: October 4, 2025
Two generations came back similarly unimpressive for this prompt
Tested: October 4, 2025
Wan 2.5 did not disappoint. A few generations came out very well, was hard to choose.
Tested: October 4, 2025
Perfect focus shift and realism, only the soldier's movements are so frantic that towards the end you can spot some warping in anatomy
Tested: October 4, 2025
Seedance Lite knows exactly what focus shift means. Underrated model.
Tested: October 6, 2025
There's an abrupt cut instead of focus shift.
Tested: October 8, 2025
Very nice focus shift. Start image is solid black.
Is the butterfly shown in extreme macro detail (texture on wings, antennae)?
Does the butterfly land on a purple flower with a white center?
Is the flower slightly off-center in the composition?
Is the background initially soft and blurred, with a greyish tone?
Do the butterfly’s wings shimmer in a way that suggests faint sunlight?
Is the camera completely steady during the macro portion of the shot?
Does a rack focus shift the background from blur to sharp clarity?
Are tanks, soldiers’ legs, and rising dust revealed clearly after the focus shift?
Is there audible or implied shouting during the background reveal?
Does the camera pull back smoothly into a wide shot of a war-torn street?
Are fire, smoke, and rubble convincingly present in the background?
Check out the results from FLOW (Veo 3 Fast) vs Kling AI (Kling 2.5 Turbo) vs PixVerse (PixVerse V5) vs Wan (Online Platform) (Wan2.5 Preview) vs Freepik (Hailuo 02) vs Dreamina AI (Seedance 1.0 Lite) vs Freepik (Sora 2) vs GROK (Grok Imagine v0.9) for similar or identical prompts side-by-side.
Martial arts sequence by zoobruh
Double dolly shot woman pointing gun
Female soldier cinematic scene
