The output preserved the source resolution exactly at 2160×3840, with no downscaling.
What was measured
Output resolution
Does it maintain the input resolution or downscale?
context, not decisivetransformation
Keeping full resolution is important for delivery quality, but it is a downstream output constraint rather than the main background-removal task itself. (3 of 3 judges)
What was given, what came back
Test input: Oregon Coast Walking · video · group: remove-or-replace-video-backgrounds-using-ai
Input — what we sent
Oregon Coast Walking
A daylight outdoor video of a single person walking along an Oregon coast beach, with the camera following the subject and ocean waves, sand, and sky visible in the background. It was used to test background removal and replacement on a moving subject in a natural outdoor setting.
Why this input is hard
- · Outdoor background removal
- · Subject segmentation during movement
- · Hair and edge preservation
- · AI background replacement quality
Output — unretouched
Also checked on this input — same tool, 2 other criteria
Edge quality⚠ StruggledA strong blue-white halo bleeds past the head and shoulders into the black background, so the silhouette is not clean at the edge.Motion handling✓ WorkedThe walking subject stayed isolated while taking a few steps and then slowing to a stop, so normal walking motion did not break the cutout.
Provenance
- Observation
- bce633a7-8281-4a69-962a-177f1820c232
- Evidence run
- 10155228-7ce8-438a-a210-547331ef080b
- Study
- Remove or Replace Video Backgrounds Using AI
- Research task
- 86ba42c2d
- Tested at
- not recorded
- Source
- first-party
- Evidence state
- verified
- Proof shown
- input + output shown
- Cost / latency
- not captured
- Repeat run
- not captured
- Tester
- not captured
The last three rows are honest blanks, not placeholders — our capture has no field for them yet.
Query this
get_evidence({
tool: "bria-ai",
scenario: "remove-or-replace-video-backgrounds-using-ai"
})MCP · mcp.aidemos.com/api/mcp
Free with attribution.
Same input, same check — 5 other tools
measured on Output resolution
Cutout.Pro⚠ StruggledDownscales the roughly 2160×3840 source to 360×640, so it does not preserve native resolution.Descript✗ FailedIt downscales a 2160×3840 source to a 720×1280 export instead of preserving the original resolution.Fotor✗ FailedA vertical 2160×3840 source was exported into a landscape 2464×1080 canvas, leaving the subject confined to a narrow center strip.InVideo AI✓ WorkedIt held the export at 2160×3838, matching the source’s roughly 4K vertical resolution instead of downscaling it.VEED◐ MixedDownscales a 2160×3840 source to 1080×1920 on export, cutting the resolution in half while keeping duration nearly unchanged (5.30s→5.27s).
This evidence is published in
Real inputs and real outputs, no retouching · every cell queryable via API & MCP · aidemos.com