01
Upload one reference image
Choose a clear character, product, object, poster, or scene. The current ClipCanva workflow accepts one reference image per generation.
Reference to Video AI Generator
Guide an AI video with one reference image. Keep a character, product, composition, or visual style closer to your source while you direct the action, camera, and pacing.
Reference to VideoCharacter ConsistencyProduct MotionSeedance
0 / 2000
Reference to Video Examples
Give the reference one clear job, then use the prompt to describe what should happen in the new shot.
Product appearance reference
Character look reference
Visual style reference
Composition reference
How to Make a Reference-Guided AI Video
Use one clear reference image, separate the visual constraints from the action, and review the result for drift.
02
Name what should stay consistent
Tell the model whether the reference should guide the face, outfit, product shape, colors, composition, lighting, or overall visual style.
03
Direct the new shot
Describe the action, camera movement, setting, mood, and pacing. Keep the motion brief specific so it does not compete with the visual constraints.
04
Review for visual drift
Check faces, clothing, product details, logos, text, hands, and backgrounds before using the clip. Regenerate with a simpler prompt when important details change.
Reference to Video vs Image to Video
Use Reference to Video for visual guidance
Choose this workflow when a character, product, composition, or visual style should influence a newly generated shot.
Use Image to Video for first-frame animation
Choose image to video when the uploaded image should become the starting frame and you mainly need to animate that composition.
Let the reference define the look
Use the image to guide identity, product appearance, colors, framing, lighting, or style. State the priority in the prompt.
Let the prompt define the motion
Describe subject action, camera direction, background behavior, atmosphere, and pacing separately from the visual reference.
Treat consistency as guidance, not a guarantee
One reference image can reduce random variation, but generated faces, products, logos, text, and small details can still drift and must be reviewed.
Choose the right video workflow
Start from a reference, a fixed first frame, or a written scene based on what you need to control.
Choose Your Plan
Free images include a watermark. Upgrade for clean outputs, faster generation, and commercial use.
Starter Pack
$9
A one-time credit pack for trying image, music, and short video generation.
- 600 credits
- ≈ 100 GPT Image 2 images at 1K Low (6 credits each)
- Up to 7 short videos
- Credits valid for 30 days
- No watermarks
Creator
$19/mo
Monthly plan · 1,500 credits renew each month
- 1,500 credits per month
- ≈ 250 GPT Image 2 images at 1K Low (6 credits each)
- Up to 18 short videos/month
- Works with ClipCanva Pro and other credit models
- Priority queue
- Community support
- No ads
- No watermarks
30 founding memberships
GPT Image 2 Unlimited
$29/mo
Use code FOUNDING6 at Stripe Checkout: $19/month for your first 6 months, then $29/month.
- Unlimited GPT Image 2 generations
- Text-to-image and image-to-image
- No credits required for GPT Image 2
- Other models use credits
- No watermarks
Best value
Pro
$39/mo
Monthly plan · 4,000 credits renew each month
- 4,000 credits per month
- ≈ 666 GPT Image 2 images at 1K Low (6 credits each)
- Up to 50 short videos/month
- Works with ClipCanva Pro and other credit models
- Highest priority queue
- Unlock ClipCanva Pro
- No ads
- No watermarks
- Advanced Refine
Studio
$109/mo
Monthly plan · 12,000 credits renew each month
- 12,000 credits per month
- ≈ 2000 GPT Image 2 images at 1K Low (6 credits each)
- Up to 150 short videos/month
- Works with ClipCanva Pro and other credit models
- Highest priority queue
- Early access to new features
- Priority support
Secure checkout · Subscriptions renew automatically · See refund eligibility
Team
For agencies, production teams, and custom volume needs.
Reference to Video AI FAQ
Choose the right input and set realistic expectations before generating.
What is reference-to-video AI?
Reference-to-video AI uses a visual reference to guide a newly generated video. The image can influence a character, product, composition, colors, lighting, or style while the prompt describes the action and camera.
What is the difference between reference to video and image to video?
Reference to video uses an image as visual guidance for a new shot. Image to video usually treats the uploaded image as the first frame and animates that existing composition. Use reference to video when the look matters; use image to video when the starting frame matters.
How many reference images can I upload?
The current ClipCanva reference-to-video workflow accepts one reference image per generation. Choose the clearest image that shows the identity, product, composition, or style you need to preserve.
What can the reference image control?
It can guide visible traits such as a face, hairstyle, outfit, product shape, colors, framing, lighting, and overall style. Tell the model which traits matter most instead of asking it to preserve everything equally.
How do I keep a character or product more consistent?
Use a sharp, well-lit reference with the subject unobstructed. Name the details that must remain close to the source, keep the action simple, and review every output for changes in faces, clothing, shape, color, labels, or logos.
What should I include in a reference-to-video prompt?
Separate the prompt into two parts: what should stay close to the reference, and what should happen in the new shot. Then add the camera move, setting, lighting, pacing, and any details the model should avoid changing.
Can I use a product photo as the reference?
Yes. A clear product photo can guide shape, materials, color, and composition. Generated video may still alter packaging, labels, logos, proportions, or small details, so review the result before publishing.
Which model does the generator use?
The page opens with Seedance 2.0 selected. Other available video models may use the reference differently, so keep the same source image and prompt when comparing results.
Create a reference-guided video
Upload one reference image, describe what should stay consistent, and direct the action in a new short video.