Collection V2: Your Characters Can Now Have Any World.
And Dialogues.
 
In the first version of the collection, each avatar had a set of stable movements against a fixed background. The idea was simple: "here's an archetype + its gestures that can be re-voiced and adapted to the task." Each has its own gesture gallery, with 10 to 25 different motions.
1. Any Background for a Familiar Gesture

The world, it turns out, knows how to surprise. Tools like Kling O1 suddenly learned to handle the task of video background replacement quite well. Instead of "removing the background" and breaking the image, you can now use the original video as a source of motion and rebuild the background from scratch, preserving the plasticity and character of the gesture.
And now, for my clients, it's possible to generate videos with avatars against (almost) any background — using the same exact gestures.
More
IRA. Collection V2
Deimos&Khalia. Collection V2
2. Dialogues: Light and Dark in One Scene

Question 1. Will this affect the avatar's character?
I don't think so. At the very least, I really wouldn't want to destroy the concept of archetypes (Light, Dark, Jokers), so the choice of character remains familiar and understandable:
  • "I need to drench the viewer in contempt — so I take Archie."
  • "If I need to politely laugh at the viewer — I take Ira..."
  • ...and so on. The logic of choice remains the same; only the scene around them can now change.
The principles of selection have already been described in previous texts, and they haven't changed.

Question 2. Will this affect the price for a video with an avatar?
Probably not. If the gesture gallery is just a showcase (and it's not final), and I'm already designing gestures for a specific client, then why not design the background too? It's the same task-oriented work. For each client — it's their own.
Therefore, I can't give a clear step-by-step plan for "how we work": there are too many nuances.
Of course, it would be easier for me to make a video with a character "as I see fit," based only on your description. But whether you'd be happy with my ideas is a big question. That's why we usually talk. For a long time. And that's why the price on the website is not a dead fixed rate (it shouldn't be!), but a starting point for discussion.
3. Syncing Two Avatars in One Shot

The evolving AI toolkit offered another pleasant surprise: you can combine two characters in one shot and lip-sync them as if in a dialogue.

This means the client, as I wrote before, can truly mix characters, not just settle on one chosen option.

For example: a conditional Dark and a Light avatar on the same background, talking to each other, not just popping into the frame one after another.

Questions about character and cost, I think, are the same here. I won't venture to say where the effort is greater: in "regular" videos with chosen gestures or in version №2 of the collection, in "v2 — backgrounds & dialogues."
Zakhra&Archi. Collection V2
In both cases, it all comes down to a conversation with the client and the scope of work. Alas (or fortunately), I am not a shoe factory churning out identical boots in 100 models and 300 colors. I simply create beautiful things to order.
4. Realistic Expectations: Magic Requires Checking
Even with the new approach, it's important to consider the reality of the tools:

  • Background replacement via different video services gives decent results in most cases, but with complex lighting or active movements, artifacts will inevitably appear somewhere — noise, "creeping" edges, color shifts.
  • Scenes with two, and especially three, characters need testing: not every gesture translates equally well into a multi-figure composition.
  • Multi-lip-sync is also not perfect: sometimes lips and sound are slightly out of sync, and this has to be reviewed and, if necessary, re-generated.
Therefore, "Avatar Collection V2" is not about "perfect magic," but about honestly expanding possibilities. New modes exist, but each specific video remains the result of careful selection of gesture, background, and tool for the task. And testing.
...we've already traveled part of the path together, which means...
5. The Result: A System of Scenes Instead of "Video Snippets"

In the end, what used to be a "collection of separate avatars" is gradually turning into a system of scenes: the characters remain the same, but the world around them becomes dynamic.

So, to new clients, I also invite previous ones: we've already traveled part of the path together, which means the next steps will be easier, faster, clearer, and... often cheaper.
Choose an Avatar from the collection
Made on
Tilda