Smoke-Test Chat: AI Video Skills Workflow Starter Kit
A representative buyer-style run showing how the final ZIP activates as a workflow assistant, asks for missing context, and produces concrete starter prompts instead of acting like a random prompt folder.
Uploads ai-video-skills-workflow-starter-kit-v1.zip into the agent workspace.
New skill activated: AI Video Skills Workflow Starter Kit.
Hi, I can help you plan a cinematic AI video workflow from your references. Let's talk about what you want to make, then I will guide the reference order, still-image prompts, Seedance/video prompt structure, and edit-bay QA before you spend generation credits.
I bought this ZIP. Help me use it to build an AI video workflow for my own project.
I can help you turn the ZIP into a working AI video workflow. I will use the included Chodigi workflow wrapper and bundled source resources. Before creating prompts, I need the minimum setup details so we do not waste generations.
- What are you making: product clip, cinematic short, character/world test, performance piece, or something else?
- Which tools are you using for still images and video?
- What runtime and aspect ratio do you need?
- What references do you already have?
- Is the subject identity already locked?
- Any rights, likeness, brand, or label constraints?
- Is generation budget tight, normal, or flexible?
I want a 10-second vertical product video for a skincare bottle. I use Midjourney for stills and Seedance for video. I have one product photo and one mood reference. Budget is low, so keep generations minimal.
Use the low-generation product route. We will preserve the bottle from the product photo, borrow atmosphere from the mood reference, generate one approved hero still, then use that still for one Seedance image-to-video pass.
Starter Artifact 1 - Reference Ordering Plan
- @image_1: product photo. Preserve bottle silhouette, cap, label placement, material, and brand-safe visible details.
- @image_2: mood reference. Use only for lighting, premium skincare atmosphere, palette, surface texture, and background softness.
- Do not copy: exact props, layout, trademarked details, or unrelated objects from the mood reference.
Starter Artifact 2 - Midjourney Hero Still Prompt
Use @image_1 as the exact skincare bottle identity reference and @image_2 only as mood and lighting reference. Create a vertical premium skincare product hero frame: the bottle centered on a clean stone vanity surface, soft morning window light, subtle water condensation, gentle cream-and-sage palette, shallow depth of field, refined editorial product photography, realistic glass/plastic material, label placement preserved from @image_1, cap shape preserved, no extra text, no distorted logo, no duplicate bottles. Composition: one bottle only, generous negative space above, elegant spa atmosphere, product remains the clear subject.
Starter Artifact 3 - Seedance Image-To-Video Prompt
Reference attachment list: 1. Approved vertical product hero still 10-second vertical skincare product video Scene & Mood: Premium skincare bottle on a clean stone vanity in soft morning light, calm spa atmosphere, subtle condensation, elegant minimal set. Frame Map: 0-3s slow push toward the bottle; 3-7s tiny light movement across the label and cap; 7-10s hold on the centered bottle with soft background blur. Subject Lock(s): The single skincare bottle keeps the same silhouette, cap shape, label placement, and material details from the reference still. Cross-Frame Rules: No duplicate bottles, no label warping, no changing cap shape, no new text, no extra hands unless provided as reference. Movement: Slow controlled product-camera push, very small parallax in background, condensation highlights shimmer subtly. Last Frame: Bottle centered and readable, premium calm finish, no sudden motion. World Plate: Clean stone vanity, soft neutral background, morning window light, gentle cream-and-sage palette. Sound Bed: optional soft room tone only if the video tool supports audio; no music instructions. Capture Realism: realistic product photography, stable label, natural material response, no plastic AI sheen. Camera Capture: vertical editorial product macro look, soft natural light, shallow depth of field, clean commercial finish.
Starter Artifact 4 - Edit-Bay QA
- Keep if bottle identity, cap, label position, and material remain stable.
- Trim if the opening or ending is good but the middle drifts.
- Regenerate only if the bottle duplicates, label melts, or motion breaks product trust.
- Before final prompt: confirm product-photo rights, readable-label requirement, exact aspect ratio, and final-frame preference.
So this ZIP is not just a prompt folder?
Correct. The prompts are included, but the product value is the workflow layer. The ZIP tells the agent how to ask for missing context, order references, route still-image work, create the video prompt, and check the output before spending more generation credits.