​

Generate videos with Krea AI

With Krea, you can now generate incredible videos from just a text prompt or starting image. Krea’s AI video models generate animations tailored for storytelling, character motion, and promotional content. Each model has different strengths in frame generation, motion consistency, and character retention.

​

Generating videos with Krea AI

Generating video clips with Krea is straightforward, especially if you’ve already used Krea to generate images. Here’s the video generation screen. In the center, you see the prompt box where you can write your text prompt, tweak additional settings, and start the generation process. On the left-hand-side, you can see previous sessions containing earlier video generations. In the bottom-left-hand corner, you can see the model picker, which lets you switch between our various video generation models. You can run up to six video generation jobs concurrently – you don’t need to wait for one video to finish before submitting a different prompt or trying a different model.

​

Video generation settings

As with our image generation tool, various settings are available, and these appear beneath the prompt box. Not all models support all settings. Aspect ratio – select from portrait or landscape dimensions.
Duration – set the length of the video, for example, 6s or 12s.
Resolution – set the video quality, for example, 720p or 1080p.
Start image and End image – set the image you want the video to start or end with. This is the key tool for turning photos (or AI images) into videos – or illustrations into animations!

​

Extending videos

Although individual video clips are relatively short – no more than 12 seconds – you can extend each video with a further generation. The AI will use the final frame of your current video (as well as the preceding video context) to generate further seconds of footage. To access this tool, hover over a previously generated image and choose Extend.

​

Adding audio

Only the Sora and Veo 3 models generate video clips with matching audio. For other models, you can choose Add audio beneath the video that you generate. A seperate AI model will then attempt to generate matching audio that complements the scene. If you don’t like the results, use the **Replace audio **button beneath the video to generate further attempts.

​

Available video models

​

Top models

These are some of our most popular video generation models, developed by a variety of providers.

Model	Creator	Length	Resolution	Compute units	Audio	Start frame	End frame	Other features
Sora 2	OpenAI	4s, 8s, 12s	720p	TBC	🔊	✅
Sora 2 Pro	OpenAI	4s, 8s, 12s	720p, 1080p	TBC	🔊	✅
Veo 3	Google	4s, 6s, 8s	720p, 1080p	3,200 ‼️	🔊	✅
Veo 3 Fast	Google	4s, 6s, 8s	720p, 1080p	1,600	🔊	✅
Seedance Pro	ByteDance	5s, 10s	720p, 1080p	3,100 ‼️	-	✅	✅	Reference images
Wan 2.5	Alibaba	5s, 10s	480p, 720p, 1080p	1,400	-	✅
Wan 2.2	Alibaba	-	480p, 720p	3,100 ‼️	-	✅
Wan 2.1	Alibaba	-	480p, 720p	1,000	-	✅	✅	Styles
Kling 2.5	Kuaishou	5s, 10s	-	1,000	-	✅
Kling 2.1	Kuaishou	5s, 10s	1080p	3,800 ‼️	-	✅	✅
Hailuo 02	Minimax	6s, 10s	512p, 768p, 1080p	750	-	✅	✅
Pika 2.2	Pika Labs	5s	1080p	1,000	-	✅	✅	Effects
Runway Gen-4	Runway	5s, 10s	720p	700	-	✅

​

Other models

These include legacy models that are no longer SOTA, but may form an important part of existing workflows. They’re available by clicking **View All Models → **in the model picker.

Model	Creator	Length	Resolution	Compute units	Audio	Start frame	End frame	Other features
Runway Gen-3	Runway	5s, 10s	-	700		✅
Seedance Lite	ByteDance	5s, 10s	480p, 720p, 1080p	650		✅	✅	Reference images
Veo 2	Google	5s, 6s, 7s, 8s	-	2,705 ‼️		✅
Hunyuan	Tencent	-	512p, 720p	1,000
Hailuo	Minimax	6s, 10s	512p, 768p, 1080p	1,000		✅	✅
Luma	Luma Labs	-	-	1,100		✅	✅
01-Live	MiniMax	-	-	1,000		✅
Ray 2	Luma Labs	-	540p, 720p	2,200 ‼️		✅
Kling 1.0 (Pro)	Kuaishou	5s, 10s	-	1,300		✅	✅

​

Animation Workflow

1. Select parameters such as aspect ratio, clip duration (1-10 seconds), motion intensity, and looping.
2. Prompt directions into the text bar, and select specific durations for the directions (e.g., “Tracking shot following the running woman, rapidly zooming into her face” for a 3 second period, and applied to a specific frame)
3. Add keyframes by uploading images, generating new images, or selecting from the library.
4. Choose a style (Film, Animation, Render, Experimental) and generate the animation.

​

Best Practices for Video Models

• Test different models with the same prompt to see which best fits the intended style.
• Use keyframes strategically—longer keyframe sequences help AI understand motion better.
• Generate frames separately before animating complex scenes to ensure consistency.
• For character-driven animations, choose models optimized for character retention like Hailuo.
• For dynamic camera movements, models like Runway offer better perspective shifts.
• Start with shorter clips (5-10 seconds) to test ideas before creating longer animations.
• Use text prompts that include specific motion descriptions for better results.

​

Creative Applications

• Brand animations for logos and visual identities
• Product demonstrations showing features and uses
• Narrative storytelling for short films or promotional videos
• Social media content optimized for platforms like Instagram and TikTok
• UI/UX animations demonstrating app or website functionality
• Educational content visualizing concepts or processes