People & characters
A selfie, a friend, a mascot or a character from a painting.
Pika 2.0 was the model that let you put your own world into AI video. Its headline feature, Scene Ingredients, let you upload photos of people, objects, outfits and places, and Pika blended them into one scene. It also followed prompts more closely and moved more naturally. Here's what Pika 2.0 brought, how it compares to today's Pika, and where its ideas live now.
Pika 2.0 was Pika Labs' next video model after Pika 1.5, arriving only weeks after 1.5 and its viral Pikaffects. Pika summed it up as superior text alignment, stunning visuals and Scene Ingredients, which let you upload images of yourself, people, places and things.
Pika pitched it squarely at everyday creators rather than big studios. Its tagline was "Not just for pros. For actual people. (Even Europeans!)", a nod to the fact that Pika 2.0 launched for all users, including those in the European Union.
It was available on pika.art and through Pika's Discord, with a free tier and paid plans that unlocked more.
| Pika 2.0 at a glance | |
|---|---|
| Made by | Pika Labs |
| Came after | Pika 1.5 |
| Headline feature | Scene Ingredients |
| Ingredients | People, characters, objects, clothing, settings |
| Other upgrades | Text alignment, visual quality, motion |
| Where | pika.art and Discord |
| Access | Free tier + paid plans, EU included |
Think of it as a pantry of visual ingredients. You upload the pieces, describe the action, and Pika works out what each image is meant to be before blending everything into one coherent clip.
A selfie, a friend, a mascot or a character from a painting.
Your cat or dog, ready to star in any scene.
A product, a surfboard, a guitar, anything you can photograph.
Dress a character in a specific outfit.
A background or location, like a night sky or a city street.
Tells Pika what happens, for example "the person is petting the cat".
Add images of the people, objects, outfits or places you want in the shot.
Write a prompt saying what the ingredients do together and how the camera moves.
The model recognises each image's role and composes a single scene from them.
Tweak the prompt or swap an ingredient, keeping your character consistent across clips.
Template clips from Pika's library in the Scene Ingredients style: a reference person, pet, outfit or object placed into a brand-new scene.
Videos © Pika, hosted on cdn.pika.art. Credit: Pika (pika.art).
Even without uploading any images, Pika 2.0 was a clear step up from 1.5:
Upload a pet photo and a background.
Upload a selfie, an outfit and a prop.
How Pika 2.0 compares with what you get on pika.art now.
| Pika 2.0 | Pika today | |
|---|---|---|
| Main model | Pika 2.0 | Pika H3 Flex, Pika 2.5 + partner models |
| Your own images | Scene Ingredients | Up to 12 references with H3 Flex, reusable characters in Character Studio |
| Clip length | Short single clips | Up to 30 s multi-shot in Video Studio |
| Sound | No native audio | Synced audio, music, speech and SFX |
| Resolution | Standard (1080p came with 2.1) | Up to 4K with H3 Flex |
| Where | Web, Discord | Web, iPhone, ChatGPT, AI agents, API |
See the full version history in Pika versions compared.
Idea-to-video on the web, with Modify Region and Expand Canvas.
Pikaffects, Big Screen Shots and more realistic movement.
Scene Ingredients, better text alignment and motion.
1080p video with sharper detail and more lifelike people.
Up to 10-second clips in 1080p and Pikaframes.
Sharper visuals and smoother motion.
Synced audio, 12 references, 4K finishing and open weights.
Pika 2.0 isn't among the featured video models in today's Pika studio, which centres on Pika H3 Flex, Pika 2.5 and partner models. The idea it introduced is now central to Pika: