This page records real outputs generated with our Enzo Celebration AI generator. It keeps the original input/output source, test record date, resolution, duration, and completeness status beside each playable file.
The library includes complete generated clips and partial shot-level outputs. A partial shot is labeled as partial; it is not presented as a complete eight-second result or as a universal success rate.
Quick answer
Every public video listed on this page was generated with Enzo Celebration AI. Complete clips from doc/resource and the prepared template outputs are kept separate from the shorter doc/spike shot tests, while every card records its input/output role, test date, dimensions, duration, and complete or partial status.
These files were generated with our Enzo Celebration AI generator and are published as real examples. The original filename stays visible in every card so the output can be traced back to the tested asset.
The files have different source dimensions and durations. Each card preserves the exact media specification instead of treating every clip as the same 720p benchmark.
Each generated file stays playable while its input/output source and test boundary remain beside the video.
Generated with Enzo Celebration AI from doc/resource/prepared/enzo-169-8s.mp4. This card preserves the original source role and the exact published output specification.
Evidence note: Complete generated output with the source filename, dimensions, frame rate, duration, and test record date preserved.
Generated with Enzo Celebration AI from doc/resource/prepared/enzo-916-8s.mp4. This card preserves the original source role and the exact published output specification.
Evidence note: Complete generated output with the source filename, dimensions, frame rate, duration, and test record date preserved.
Generated with Enzo Celebration AI from doc/resource/示例1.mp4. This card preserves the original source role and the exact published output specification.
Evidence note: Complete generated output with the source filename, dimensions, frame rate, duration, and test record date preserved.
Generated with Enzo Celebration AI from doc/resource/示例2.mp4. This card preserves the original source role and the exact published output specification.
Evidence note: Complete generated output with the source filename, dimensions, frame rate, duration, and test record date preserved.
Generated with Enzo Celebration AI from doc/resource/示例3.mp4. This card preserves the original source role and the exact published output specification.
Evidence note: Complete generated output with the source filename, dimensions, frame rate, duration, and test record date preserved.
Generated with Enzo Celebration AI from doc/resource/示例组1.mp4. This card preserves the original source role and the exact published output specification.
Evidence note: Complete generated output with the source filename, dimensions, frame rate, duration, and test record date preserved.
Generated with Enzo Celebration AI from doc/resource/示例组2.mp4. This card preserves the original source role and the exact published output specification.
Evidence note: Complete generated output with the source filename, dimensions, frame rate, duration, and test record date preserved.
Generated with Enzo Celebration AI from doc/resource/示例组3.mp4. This card preserves the original source role and the exact published output specification.
Evidence note: Complete generated output with the source filename, dimensions, frame rate, duration, and test record date preserved.
Every file in this section is a first-party generated output. Complete clips and partial shot tests are kept in separate sections and are not combined into a success-rate denominator.
The three entries below pair an input template shot with the corresponding output generated by our Enzo Celebration AI shot-replacement pipeline. They are genuine first-party test outputs, but each output covers only part of the full sequence.
The output files are 1408×768 at 24 fps and run for 2.00 or 3.58 seconds. They are useful for inspecting replacement behaviour and motion continuity; they do not establish the quality of a complete eight-second product render.
Each card keeps the input and output beside the exact test boundary.
Input and output from doc/spike/shots-t1-enzo-169. The output is a partial shot, not a complete final video.
Evidence note: This is a shot-level output. It must not be counted as a passed eight-second render.
Input and output from doc/spike/shots-t2-enzo-169. The output is longer than t1 but still covers only one partial sequence.
Evidence note: This output is evidence of one experiment, not a population-level quality metric.
Input and output from doc/spike/shots-t3-enzo-169. The output is retained as a separate test run so the result is not confused with t2.
Evidence note: No success rate is calculated from t1–t3 because the sample is partial and the acceptance matrix is not complete.
These are first-party generated shot tests. Their partial status is explicit so they are not mistaken for complete product renders.
The same checks will be used when full product renders are added.
Does the replacement remain recognisable through the visible shot?
Which declared movement is actually visible in the available clip?
Are hands, fingers, ears and the face readable where they overlap?
Does the visible clothing and frame match the test setup?
Do subject position, clothing and background remain stable between frames?
Audio is recorded only when it exists in the published file; silent files remain labeled silent.
No aggregate score or success rate is published until the sample, acceptance rule, and denominator are fixed.
This library proves that the listed files are real first-party generation runs. It does not prove that every future uploaded photo will preserve identity, clothing, hands, or framing, and it does not establish a universal success rate or superiority over another tool.
A future full-render report will use a declared input set, one generation per input, a pass/partial/fail rule, the complete output set, and a dated results table. Until then, the honest label for an unmeasured claim is Not stated.
Got questions about video credits, payment safety, or export quality? Find direct answers below.
doc/spike. Each output is partial, so it is labeled as a shot-level test rather than a complete eight-second result.Upload one consenting adult's photo, review the photo check, and see the exact price before rendering.
Open the generator →Written and maintained by the Enzo Celebration AI editorial teamPublished Last updated