Pika H3 Flex Image to Video
Upload a start frame — and optionally an end frame — and Pika H3 Flex animates it into a 5–15 second clip with synchronized stereo audio. No GPU, no ComfyUI setup: it runs in your browser.
- Start frame, end frame or both
- Synced stereo audio
- 5, 10 or 15 seconds
- 480p → 4K
- Credits back if it fails
Duration
Resolution
Aspect ratio
No free tier · credits from $9.90 · failed videos refunded · Pricing
Prompts are screened by automated moderation · Acceptable use
Cinematic night street in the rain
Night street in a rain-soaked city, neon signs reflecting in puddles, a woman in a yellow raincoat walks toward camera, slow dolly back, shallow depth of field. Sound: steady rain, distant traffic, soft synth pad.
Looking for the other mode? Pika H3 Flex text to video
How to turn an image into a video with Pika H3 Flex
- 1
Upload a start frame
Drop in a JPEG, PNG or WebP (up to 10 MB). This becomes the first frame of the clip, so use the exact composition you want to open on.
- 2
Add an end frame (optional)
Give a second image and the model bridges the two — a first/last-frame transition. Leave it empty to let the shot evolve freely from the start frame.
- 3
Describe motion and sound
The image already shows what is in the shot. Use the prompt for what changes: camera move, action, and the audio you want to hear.
- 4
Pick length and resolution
Choose 5, 10 or 15 seconds and 480p to 4K. The exact credit cost shows on the Generate button before you start.
Real outputs
Real Pika H3 Flex outputs, prompts included
Every clip below was generated with Pika H3 Flex on this site from the exact prompt shown — text to video, 768p, 10 seconds, 40 credits each, with the sound generated in the same pass. Hover to preview, click to play with audio. Web previews are compressed.
Cinematic night street in the rain
Night street in a rain-soaked city, neon signs reflecting in puddles, a woman in a yellow raincoat walks toward camera, slow dolly back, shallow depth of field. Sound: steady rain, distant traffic, soft synth pad.
Vertical social clip: skateboarding cat
Vertical 9:16, a cat wearing tiny sunglasses rides a skateboard across a polished kitchen floor, low tracking shot. Sound: skateboard wheels on tile, surprised gasp, punchy beat drop.
Product ad: wireless earbuds
A matte black wireless earbud case opens on a wet stone pedestal, water droplets bounce in slow motion, studio rim light, slow macro orbit. Sound: crisp case click, water drops, deep bass hit.
Dialogue: the old fisherman
Close-up of an elderly fisherman on a wooden dock at golden hour, he looks into the lens and says "The sea always gives back what it takes." Sound: his calm voice, gulls, gentle waves.
Food B-roll: iced coffee pour
Iced coffee poured over clear ice in a tall glass, condensation running down the glass, slow motion, warm morning window light. Sound: ice cracking, pouring liquid, light acoustic guitar.
Drone reveal over an alpine lake
Drone rises over a misty pine ridge at sunrise to reveal a turquoise alpine lake, smooth slow ascent. Sound: wind, distant birds, swelling orchestral strings.
Live music: jazz club
A jazz trio plays in a smoky basement club, slow push-in on the saxophonist, warm tungsten light. Sound: live saxophone solo, upright bass, quiet crowd murmur.
90s sitcom gag
Bright 90s sitcom living room, a grandpa tries to take a selfie with a flip phone and the flash goes off in his eyes, static wide shot. Sound: studio audience laughter, camera shutter, sitcom music sting.
Tips for better image-to-video results
Match the aspect ratio to your image
Pick the aspect ratio closest to the uploaded frame (16:9, 9:16, 1:1, 4:3, 3:4 or 21:9) so the start frame is not cropped or padded.
Prompt the change, not the picture
“Slow push-in, steam rises from the cup, quiet café chatter and a spoon clinking” works better than re-describing everything that is already visible.
Keep end frames plausible
First/last-frame clips look best when both images share the subject, lighting and lens. Two unrelated photos force a hard morph.
Draft cheap, finish sharp
Test motion at 480p or 768p, then re-run the version you like at 1080p, 2K or 4K. Higher resolutions are upscaled from a 768p generation.
Ask for the sound you want
Audio is generated in the same pass as the picture. Name dialogue, music, sound effects or ambience in the prompt to steer it.
Image-to-video specs on this site
| Inputs | Start frame, end frame, or both + text prompt |
|---|---|
| Image formats | JPEG, PNG, WebP, up to 10 MB each |
| Clip length | 5, 10 or 15 seconds at 24 fps |
| Resolution | 480p, 768p (native); 1080p, 2K, 4K (upscaled) |
| Aspect ratios | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 |
| Audio | Synchronized stereo, generated with the video |
| Output | MP4 download, no watermark, commercial use |
Pika H3 Flex also has a multimodal reference-to-video mode (up to 12 reference images, videos and audio clips). It is available through Pika’s API and the open weights; this studio currently supports start/end-frame image to video.
Related guides
Transparent costs
Pika H3 Flex credit costs per video
Credits are charged per second of output. The table shows the full cost of one clip; failed generations are refunded.
| Resolution | Credits / second | 5s clip | 10s clip | 15s clip |
|---|---|---|---|---|
| 480p | 2 | 10 | 20 | 30 |
| 768p | 4 | 20 | 40 | 60 |
| 1080p(upscaled) | 7 | 35 | 70 | 105 |
| 2K(upscaled) | 8 | 40 | 80 | 120 |
| 4K(upscaled) | 16 | 80 | 160 | 240 |
Pika H3 Flex Image to Video: FAQ
Can Pika H3 Flex turn a photo into a video?
+
Yes. Upload the photo as the start frame, describe the motion and sound you want, and the model generates a 5–15 second clip that opens on your image, with synchronized stereo audio.
What is first/last-frame image to video?
+
You give two images — where the shot starts and where it ends — and Pika H3 Flex generates the motion in between. It is useful for transitions, product reveals and before/after shots.
Do I need a prompt if I upload an image?
+
A prompt is required, but it can be short. Focus on what should move, how the camera behaves and what we should hear; the image already defines the subject and style.
How many credits does an image-to-video clip cost?
+
The same as text to video: credits per second of output by resolution. A 5-second 768p clip costs 20 credits, a 10-second 1080p clip costs 70. Failed generations are refunded automatically.
Can I run Pika H3 Flex image to video locally instead?
+
Yes, the weights are open (first-frame, last-frame and first/last-frame tasks), but local runs need a datacenter GPU — Pika tested on an 80 GB H100 — and about 216 GB of model files, and output stops at 768p. See our local setup guide.