Zol AI Lab
← All projects
Product · since October 2026prototype

ZolStudio

One AI studio for creators, podcasters and advertisers.

A connected generative-AI studio where content creators, podcast owners and TV and online advertisers produce images, video and talking shots in one place, running the models on our own GPU.

Content creatorsPodcast ownersTV and online advertisers
4studios in one app
3generative models
2Knative image resolution

How it works

From input to finished result.

1
Your ideaWords, a photo, a voice track or an existing clip
2
ZolStudioPicks the recipe, queues the job, keeps your files private
3
AI enginesLTX 2.5, MiniMax H3 and Qwen-Image 2.1 on ComfyUI
4
NVIDIA GB10Runs on our own GPU, with safety guards
5
Your libraryFinished images and videos, ready to publish

What it does

The parts that make ZolStudio useful.

Image

Pictures from words at native 2K, or edits that combine up to 10 reference pictures.

Qwen-Image 2.1

Video

Video with sound from words or from a start frame; MiniMax H3 can also aim for an end frame.

LTX 2.5 · MiniMax H3

Lip Sync

A portrait and a voice track become a natural talking shot, for hosts, ads and explainers.

MiniMax H3

Trends

One-click Character Swap and Cast Swap: put new people into an existing clip, with face tracking, speaker detection, a photo quality gate and a face check of the result.

MiniMax H3 · Qwen-Image 2.1

Models and tools

What ZolStudio runs on.

The interface builds on the open-source Open Generative AI project (MIT licence).

Names and logos belong to their owners and identify the models and tools we use; no endorsement is implied.

Built so far

Engineering that is already in place.

  • A fair queue runs one GPU job at a time; jobs survive a closed tab
  • Every file belongs to a user and is checked on every read
  • Engines are reached only through their API, so they can move to cloud GPUs
  • Final and Draft quality recipes for each model
  • English and Arabic, right to left

Where it's going

The next steps, in order.

  1. Ready for the publicCommercial model licences, sign-in shared with ZolVoice, content safety and consent for real faces, a visible AI mark
  2. Ready for many usersMore GPU servers as engines, credits and fair quotas, a database and cloud storage
  3. More trendsOutfit Swap, One Ad Every Market and Dance Trend

Lab notes

Progress on ZolStudio.

ZolStudio: first version

The first version of ZolStudio runs image, video and lip-sync models on our own GPU, behind one simple studio: pictures from words, video with sound, talking shots from a portrait and a voice, and one-click character swaps in existing clips.

product · generative AI

Designed and built at Zol AI Lab by Moe Eldouma, founder. See his role, skills and other work on eldouma.dev.