How to Use Create from Scratch (Image to Video) for Video Creation

Bring your still images to life with smooth, controlled motion, and use end frames to guide exactly how each clip finishes. You can also combine multiple clips to create one cohesive video.

Create from Scratch gives you control over every storyboard frame, making it ideal for garments that may be harder to generate with Voodoo Video due to safety restrictions.

What Create from Scratch legacy (Image to Video) Tool does

Caimera's VooDoo tool has a full agent for generating campaign and marketing videos automatically. But when you want hands-on control over a single clip — the exact pose, the exact movement, the exact ending — Create Videos from Scratch (our image-to-video tool) is the one to reach for.
You give it a start frame and, optionally, an end frame and a short movement prompt. It animates the motion between them. You can also chain several clips together to build a video longer than one clip's limit, all from the same panel.

Step 1: From your Caimera home screen, look at the left-hand sidebar and click Video – VooDoo under the Marketing section.

Video – VooDoo also lives in the left-hand navigation panel.

Step 2: You'll land on the video agent's home screen. This is where you'd normally go for the full campaign-video agent or UGC-style videos — but for manual, frame-by-frame control, click the Create Videos from Scratch card (marked Legacy). This is our original image-to-video tool.

Choose "Create Videos from Scratch" instead of the marketing agent or UGC templates.

Step 3: The Image to Video panel opens with clips on the left and your generation history on the right. Click the Start Frame box and upload the image you want the video to open on — any file from your device or your Caimera library works.

Every clip needs a start frame before you can do anything else.

Step 4: In the Video prompt field, describe only the desired movement (e.g., how the model poses, turns, or walks) along with any audio or mood preferences. Do not describe the outfit or model's appearance, as the tool already extracts those details from your start image.

"Model posing standing with hand on hip, country music" is a good example: pure movement and mood, no styling detail.

Step 5: Add an end frame (optional) if you want the model to turn from front to back, or move from one pose into another. Click End Frame and upload the pose you want the clip to finish on. Caimera animates the motion between your start and end frame automatically.

Skip this if you just want the model to move naturally with no fixed ending pose.

Stacking clips for a longer video

When using the Kling 2.6 Pro and Pro models, the maximum clip length is 5 seconds.

If you want to create a longer, cohesive video, we recommend structuring your clips and storyboard as multiple shorter clips and combining them into a complete sequence. This gives you more control over the flow and helps maintain consistency throughout the video.

Step 6: Click Add New Clip below your first clip to add another start-and-end-frame block. Repeat as many times as you need — you can drag clips to reorder them, and each one adds its own duration and credit cost to the final video.

Each new clip gets its own start frame, end frame, and prompt.

Choosing your model

The model you pick decides your credit cost, whether you can generate audio, and whether you can set a custom clip duration or you're stuck stacking 5-second clips.

Step 7: Open the Select Model dropdown and pick from Kling 2.6 Pro, Pro, or Kling 3.0 Pro depending on the budget, audio, and duration you need (see the table above).

Each model shows its starting credit cost right in the dropdown.

Step 8: If you chose Kling 3.0 Pro, a Duration (seconds) dropdown appears. Pick anywhere from 10 to 15 seconds — you're charged based on the duration you select, so shorter clips cost less.

Longer duration, higher credit cost — pick only what the shot needs.

Step 9: Check Generate audio if you want sound in your clip (Kling 3.0 Pro only). Enhance prompt is on by default and rewrites short prompts for richer detail — switch it off if you've already written a long, specific prompt yourself.

Both settings sit just below the model and duration selectors.

Step 10: Once every clip has its frames and prompt set, click the Generate button at the bottom of the panel. The credit cost shown updates live as you add clips, change models, or adjust duration.

Videos typically take a few minutes to render.

Step 11: Find your video in the history panel. You'll see an "Added to Queue" confirmation, and a new tile marked in_progress appears at the top of the History panel on the right. Once it's ready, click it to preview, download, or reopen it any time to check the prompt you used or generate a variation.

Every video you generate is saved here, so nothing is ever lost.

FAQs

1. What's the difference between this tool and the video marketing agent?

The marketing agent and UGC templates automate an entire campaign video for you. Create Videos from Scratch is the manual, image-to-video tool; you control the exact start frame, end frame, prompt, model, and duration for every clip yourself.

2. Can I add new elements to the video, like props or a background change?

No, this tool only adds motion to the image you upload. If you need to change something in the shot itself, edit the image first (for example with Upload & Edit) and bring the finished image in as your start or end frame.

3. What should I actually write in the video prompt?

Just the movement: how the model poses, turns, or walks, and any mood or audio cue you want. Don't describe the outfit, the model's looks, or anything already visible in the image the tool already knows what's in your photo, so that part of the prompt is wasted.

4. How long can one video be?

A single clip on Kling 2.6 Pro or Pro is a fixed 5 seconds. Kling 3.0 Pro lets you choose anywhere from 10 to 15 seconds in one clip. For anything longer, stack multiple clips together 'Add New Clip'—there's no hard limit on how many you can chain.

5. How do I make stacked clips look like one continuous video instead of separate cuts?

Match the end frame of one clip to the start frame of the next. If the pose and framing line up exactly across the join, the clips play back as one smooth, cohesive video rather than a hard cut. See the "Stacking clips for a longer video" section above.

6. Why don't I see a duration option or audio toggle for some models?

Kling 2.6 Pro and Pro are fixed at 5 seconds per clip with no audio generation, they're built for stacking cheaply. Only Kling 3.0 Pro supports a custom 10–15 second duration and audio generation, billed based on the duration you choose.

7. How many credits will a video cost me?

It depends on your model and clip count: Kling 2.6 Pro starts at 6 credits per clip, Pro at 10 credits per clip, and Kling 3.0 Pro from 9 credits, scaling with the duration you pick. The Generate button always shows the live total before you commit.

Related Articles:

🎬 How to Edit Your Video Generations in Caimera: Trim, sequence, and manage your full timeline

🎞️ How to Add Existing Clips to Your Video Generations: Pull from your library to build multi-scene videos

🔍 How to Fine-Tune Frames and Prompts for Specific Clips: Dial in individual clip quality before adding overlays

🎬 Prompting on Move: How to Use Video Instructions and get stronger results with targeted motion prompts

Have a question or running into an issue? Reach us via the chat bubble or at contact@caimera.ai.

Updated on: 11/09/2026