Multi-Format Export Workflow: One Brief to All Platforms
The 4-step multi format ai export workflow that generates natively per aspect ratio, locks identity across formats, adds brand layers, and schedules in 32 minutes for $7.40.
A multi format ai export workflow generates once per aspect ratio natively, then locks identity and brand layer across every format before publishing. One brief in. Four platform-ready outputs out. For a standard social drop covering 1:1 PDP, 9:16 Reel, 16:9 YouTube, and 4:5 feed, the chain runs in 32 minutes at $7.40 in model credits.
TL;DR
- Generate per aspect ratio natively. Cropping a 16:9 source to 9:16 degrades quality and loses key elements. The workflow routes the same brief to four separate generation jobs, one per format.
- Identity locking via Higgsfield Soul 2.0 or Seedance 2.0 reference conditioning keeps the subject, product, and color treatment consistent across outputs.
- Brand layer per format means safe zones, logo placement, and text overlays are specified per platform spec, not applied universally.
- The full four-format run for a social launch asset takes 32 minutes and costs $7.40 in generation credits at current 8frame rates.
Why re-cropping kills quality
Generating one 16:9 master and cropping it into other formats is faster in theory. In practice, three things go wrong.
Composition breaks. A 16:9 frame is composed for a horizontal layout. Crop it to 9:16 and you cut off the subject's shoulders, lose the product in frame, or introduce empty headroom. The composition was never designed for that container.
Text safe zones shift. A 9:16 Reel has a caption strip across the bottom 20% and a right-side action bar. A 1:1 grid post has center-weight rules. Text placed for 16:9 lands under platform chrome or outside the crop boundary in every other format.
Motion direction reverses. A left-to-right pan that reads naturally across a wide 16:9 field can feel cramped or cut short in 9:16. Generated motion needs to be directed for its container from the start.
Generating each format natively takes more queue time. It produces outputs that look designed, not repurposed.
The 4-step multi-format export workflow
Step 1: Generate per aspect ratio natively
The workflow splits one brief into four parallel generation jobs. Each job passes the same subject, style direction, and key message to the model, but specifies its own aspect ratio and composition guidance.
The brief structure for a social product launch:
Product: [product name + 1-line description]
Subject: [what the video shows: person, product, or scene]
Style: [e.g., bright studio, lifestyle outdoor, clean editorial]
Key message: [the one thing the viewer should take away]
Duration: [per format, see table below]
The four format jobs this brief feeds:
| Format | Aspect ratio | Duration | Model | Composition note |
|---|---|---|---|---|
| PDP (product detail page) | 1:1 | 6-8 sec | Seedance 2.0 | Product centered, clean background |
| Instagram Reel | 9:16 | 15-30 sec | Kling 3.0 | Subject upper third, safe zone bottom 25% |
| YouTube pre-roll | 16:9 | 15-30 sec | Kling 3.0 | Full-width composition, horizontal motion |
| Instagram feed | 4:5 | 6-15 sec | Seedance 2.0 | Near-square framing, product prominent |
What each generation costs at current 8frame rates:
- 1:1 PDP via Seedance 2.0 (8 sec): $0.80
- 9:16 Reel via Kling 3.0 (30 sec): $2.40
- 16:9 YouTube via Kling 3.0 (30 sec): $2.40
- 4:5 feed via Seedance 2.0 (10 sec): $1.00
- Reference still via Nano Banana Pro: $0.80
Total generation cost: $7.40. These four jobs run in parallel on 8frame. All four are queued simultaneously and the longest job (Kling 3.0 at 30 seconds) takes approximately 90 seconds. Total queue time with parallelization: under 4 minutes for all four raw clips.
Step 2: Identity-lock the subject across formats
Four separately generated clips will drift. The person's face will vary slightly, the product color will shift between the Seedance and Kling outputs, and the lighting direction can reverse between the 16:9 and 9:16 versions. That drift is invisible in isolation and obvious the moment a viewer sees both the Reel and the feed post side by side.
The fix is reference conditioning before generation, not post-production compositing after.
For human subjects: Supply a 3-image reference pack to Higgsfield Soul 2.0 before running the Kling and Seedance jobs. Include face-forward, three-quarter, and slight elevation angles. In a three-format test (1:1, 9:16, 4:5) with a brand founder as subject, face geometry matched across all three clips without fix-up. Generation time per clip at 1080p: 75-90 seconds.
For product subjects: Use Nano Banana Pro to generate a canonical product still. Feed it as reference into every downstream Seedance and Kling job. This prevents the most common failure: Kling slightly desaturating a brand-red product that Seedance rendered at full chroma.
Tested reference-lock prompt for a skincare product:
[Reference: Nano Banana Pro still, rose-tinted glass bottle, silver cap, white studio background]
Product centered, soft shadow, slight upward camera angle, white or gradient background,
consistent color with reference, no label blur, [aspect ratio specified per job], 8 seconds
What this produced across the 1:1 and 4:5 jobs: product color matched within perceptible tolerance, label text stayed legible in both frames, shadow direction was consistent. The 9:16 Kling job with the same reference had one minor cap color shift that a prompt tightening pass resolved in the next generation.
Step 3: Apply brand layer per format
Brand layer is everything that wraps the raw clip: text overlays, logo placement, end cards, safe zone compliance. The mistake is applying a universal brand layer and then cropping it per format. The correct sequence is to specify the brand layer per format, matched to that platform's actual chrome and safe zones.
Platform-specific brand layer specs:
1:1 PDP: Logo upper-left. No text in the bottom 10% (platform attribution row). End card is not typical for PDP embeds.
9:16 Reel/Shorts: Logo upper-left. No text in the bottom 20% (caption strip) or right side 15% (action bar). CTA in the middle-lower safe zone, 30-50% from the bottom.
16:9 YouTube pre-roll: CTA in lower-third at second 5, not before. The first 5 seconds need to carry without relying on text (skip button).
4:5 feed: Grid thumbnail is the center 1:1 crop of the 4:5 frame. Keep nothing critical in the lateral 10% of the frame.
In the 8frame canvas, the brand layer is a node that runs after generation and before export. Set the spec once per format, save it to the workflow, and it applies on every future run.
Step 4: Schedule and publish
With four formatted clips out of the workflow, the output formatter node on 8frame exports each with the correct codec (H.264), resolution, and filename tag: product-pdp for 1:1, reel-9x16 for vertical, yt-16x9 for YouTube, feed-4x5 for the feed post.
Connection to a scheduling tool (Buffer, Later, or your platform's native scheduler) is external to 8frame. Export to your connected storage folder and your scheduler picks them up automatically if you've configured folder-based import rules.
Full walkthrough: one brief to four formats in 32 minutes for $7.40
Skincare product. Rose-tinted glass bottle, silver cap, no talent, clean editorial studio.
Reference still (Nano Banana Pro): 85 seconds. One 4K still used as reference for all four jobs.
Parallel generation (all four queued simultaneously):
- 1:1 Seedance 2.0, 8 sec: 78 seconds
- 9:16 Kling 3.0, 30 sec: 92 seconds
- 16:9 Kling 3.0, 30 sec: 88 seconds
- 4:5 Seedance 2.0, 10 sec: 72 seconds
Longest job: 92 seconds. All four clips ready at the 92-second mark.
One fix-up: 9:16 Kling clip had the product sitting too low, inside the caption safe zone. Regenerated with an adjusted vertical composition prompt. Add 90 seconds.
Brand layer node (first-time setup): 8 minutes. Under 1 minute on every subsequent run.
Export and rename: 4 minutes.
Total (first run): 32 minutes. Second run: 12-14 minutes. Generation cost: $7.40.
Pitfalls to watch for
Text safe zones. Instagram's bottom 25% on a Reel disappears under the caption strip and action bar. Build safe zone compliance into the brand layer node, not a manual review step.
Motion direction in vertical formats. A left-to-right pan in 9:16 crosses a narrow frame and can feel rushed or exit the subject early. Add "gradual movement, subject stays in frame throughout" when generating horizontal motion for vertical containers.
Key element clipping. Compose the reference still for the tightest format (1:1). A product label sitting in the right 15% of a 1:1 frame will clip in 9:16 and 4:5 crops. Wider formats give you extra breathing room. Tighter ones don't forgive off-center placement.
FAQ
Does this workflow work with human subjects or only product shots?
Both. For human subjects, substitute Nano Banana Pro reference conditioning with a Higgsfield Soul 2.0 multi-reference pack (3-5 angles). The identity lock step becomes more critical since face geometry drifts across separate Kling and Seedance generations in ways that product color drift does not. Plan for one fix-up call per format when running human-subject content for the first time with a new character reference.
Can I run more than four formats in the same workflow?
Yes. Add a job node for each additional format. Common additions are 2:3 for Pinterest, 1:91:1 for LinkedIn, and a 16:9 square-safe 1080x1080 for platforms that crop 16:9 to 1:1 in grid view. Each additional format adds one generation job at the relevant model's credit rate. Parallel execution keeps wall-clock time close to the longest single job.
What if my brand uses a specific color that models keep getting wrong?
Name the color explicitly and add a hex reference in the prompt when the model supports it, or include "match color exactly to reference image, no saturation shift" in the reference conditioning step. Seedance 2.0 holds product color better than Kling 3.0 when given a strong visual reference. For brand-critical color accuracy, use Seedance for the product-prominent formats (1:1 PDP, 4:5 feed) and Kling for the motion-heavy format (9:16 Reel) where product is secondary to movement.
For more on the 9:16 vertical format specifically, see Kling 3.0 prompts for Instagram Reels for composition and motion prompts tested at this aspect ratio. The multi-format export workflow template is available to clone from the 8frame workflow library.