Strata · Video generation · 2026-11

Video generation with Strata: an ad made only with local AI, reviewed and fixed by Qwen

Using only the two PCs on a LAN from the Strata research, we made a 15-second ad for a fictional ketchup, “TOMARO”. Strata (Qwen3.8-Flash-Next) does the planning, the review, the fix plans and the choice of takes; dedicated models make the images, video and voice. No cloud. We also tested whether Qwen can look at the finished video, point out problems by itself, and fix them.

Decisions
Strata 0.2.15 (Qwen3.8-Flash-Next)
Images
Qwen-Image 2.1 Turbo
Video and sound
MiniMax-H3 (stable-diffusion.cpp)
Narration
VOICEVOX (No.7)
Time per ad
about 25 min (two PCs in parallel)
Setup

Setup: only two PCs on a LAN

The two PCs are wired to the same LAN (1 Gb/s). No internet cloud is used. Both are the machines from the Strata research and physical AI pages. Video takes the longest, so the cuts are split across the two GPUs and made at the same time.

PCCPURAMGPURuns
PC 1 (decisions and finishing)Core i7-7800X (6 cores, AVX-512)125 GiBRTX 5060 Ti 16 GBStrata (planning, review, fix plans, choosing takes), Qwen-Image (images), MiniMax-H3 (video, 3 cuts), VOICEVOX (CPU), ffmpeg, the record database
PC 2 (video)Xeon E5-2630 v2 × 2 (AVX only, 2013 generation)251 GiBRTX 3060 12 GBMiniMax-H3 (video, 2 cuts), with stable-diffusion.cpp rebuilt for an AVX-only CPU
Tools

Tools: the AI that decides is separate from the AIs that make

ToolRoleSettings / time
Strata 0.2.15 + Qwen3.8-Flash-Next (NVFP4)Planning and shot list, reviewing the finished video (reads images), fix plans, choosing takes; all with thinkingplan 195 s, review 4–12 min, fix plan 6–10 min
Qwen-Image 2.1 Turbo (Q8_0)The first frame of each cut and reference pictures of people and products; can draw while looking at reference pictures8 steps, 832×480, 15–22 s per image
MiniMax-H3 (ref2va pruned Q4_K_M, stable-diffusion.cpp)Makes about 3 s of video plus matching sound effects from the first frame73 frames (24 fps), 20 steps. PC 1 about 400 s, 15.2 GB peak; PC 2 about 575 s, 11.3 GB
VOICEVOX 0.26.1 (CPU)Narration per cut (speaker No.7, announcer style)about 3 s for five lines
ffmpeg・PillowJoining cuts, captions, mixing sound effects and narrationa few seconds
SQLiteStoryboard database: shared elements (people, products), versions of each cut, review findings, fix plans, timings—

DeepSeek (V4.1 Flash) also runs on Strata, but it is text-only and cannot read images, so it was not used for reviewing the video.

Roles

Roles: Strata decides, the tools make, fixed rules connect them

A person only gives the brief and does not step in afterwards. Strata answers in JSON, and fixed code passes it straight to the tools.

  1. StrataPlan and shot list — A fictional brand, selling point and catchphrase; for five 3-second cuts, the image, motion and sound instructions (English) plus narration and captions (Japanese).
  2. Qwen-ImageFirst frames — One per cut.
  3. MiniMax-H3Video and sound effects — Split across the two PCs, at the same time.
  4. VOICEVOXNarration — On the CPU while the video is made.
  5. Fixed codeFinishing — Trim each cut to 3.0 s, join, add captions, mix narration over the sound effects.
  6. StrataReview — Looks at three frames per cut (start, middle, end) and the plan, and lists problems, severity and how to fix them.
  7. StrataFix plan — Registers shared elements (people, products), decides keep/redo per cut with new instructions, picks the caption style, and declares what cannot be fixed.
  8. StrataChoosing takes — From every version kept in the database, picks the best take for each cut and assembles them.
First ad

The first ad: done in about 25 min, but with problems such as an empty bottle and a different boy

The brief: “a typical modern 15-second ketchup ad, for a fictional product”. Strata invented “TOMARO Sweet-and-Rich Ketchup” (catchphrase “sweet and rich, nice and thick”, aimed at parents who want brighter family meals) and five cuts: family table → pouring on scrambled eggs → ripe tomato → the boy eats → product shot. Below are the first frames of each cut (Qwen-Image).

StepToolTime
Plan and shot listStrata194.7 s
Five first framesQwen-Image81.9 s
Video, five cuts (two PCs in parallel)MiniMax-H319.9 min
Five narration lines (parallel to video)VOICEVOX6 s
Finishingffmpeg1.4 s
Total (wall clock)about 25 min
Review by Qwen

Review by Qwen: it found the big problems by itself, even without a checklist

We showed Strata three frames per cut (15 images) from the finished ad plus the plan, and asked it to point out problems, without telling it the answers. A had no checklist; B had eight generic checks that fit any ad (product look, consistency of people, things not in the instructions, broken shapes, motion, text, sound and structure, the ad as a whole).

Problem (found by a person)A: no checklistB: generic checks
The boy differs between cuts 1 and 4◎ major◎ major
The product bottle is empty◎ major◎ major
Hands appear in the product shot○ medium◎ major (and the broken fingers)
Captions△ inconsistent style○ unreadable against the background
No music××
Time611 s238 s
Self-fixing

Self-fixing: shared reference pictures removed the big problems; picking from all takes gave the best result

We added a storyboard database (SQLite) and kept people and products that appear in several cuts as “shared elements”: a description plus a reference picture, stored once. Strata reads the review findings and writes a fix plan, code remakes the cuts accordingly, and Strata reviews again. Every version of every cut, the finding behind it and the fix are kept.

RoundWhat Strata didMajorMediumMinorTime
0First ad452~25 min
1Registered the boy, mother and bottle as shared elements, remade 4 cuts, bold captions on a band163~44 min
2Remade all 5 cuts, outlined captions; declared “no tool to add background music” as unfixable243~44 min
3No remaking: picked the best of all 14 takes in the database for each cut134~19 min
Major / medium / minor are the counts of Strata’s review findings on that round’s finished ad.
Finds

Big problems, even without a checklist

Including several a person missed. It cannot hear audio.

Fixes

Shared references aligned people and products

Music, with no tool for it, was declared unfixable.

Picks

Choosing the best of all takes worked best

About 19 min with no remaking, possible because every take is kept.

Notes
Why the videos are not shown

The videos were made with MiniMax-H3, whose MiniMax H3 Community License excludes the EU, the UK, the Republic of Korea and the USA and does not allow outputs to be displayed there. Because this page can be viewed worldwide, the videos and frames from them are not shown. Every picture on this page was drawn by Qwen-Image 2.1 (Apache-2.0).

All pictures, products, people and the brand on this page are fictional and machine-generated. Narration voice: VOICEVOX:No.7.