Contact person: AI Video Generation Platform Development

Phone:Show

Email:Show

Location: Hamburg, Germany

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"Full-Stack Developer Team for Hasaballa AI Video Generation Platfo
## Project Overview
We are seeking a skilled and dedicated team of developers to build **Hasaballa AI**, an **offline-first**, AI-powered video generation platform designed for **Arabic-speaking creators**.

The platform leverages **advanced AI models** to generate ultra-realistic images, voices, animations, and full videos (1-15 minutes) based on user scripts, supporting **all Arabic dialects** with a **100% Arabic RTL interface**.

Hasaballa AI operates entirely offline (**no internet, cloud, credits, or subscriptions**) on a high-performance system:

- Windows 11 Home 64-bit
- AMD Ryzen™ 9 9955HX
- RTX 5090 24GB GDDR7
- 64GB DDR5 RAM
- 4TB M.2 NVMe SSD
- 16.0" WQXGA 300Hz display

The platform aims to produce cinematic videos **matching or exceeding the quality of the following examples** (please review all links for minimum quality standards):
- [login to view URL]
- [login to view URL]
- [login to view URL]
- [login to view URL]
- [login to view URL]
- [login to view URL]

---

## How Hasaballa AI Works
Hasaballa AI offers three modes of operation for maximum flexibility:

1. **Smart Director – Automatic Mode**
Users write a full script in the Hasaballa GPT chat window (scenes, images, voices, animations), click the "Smart Director – Auto" button, and the system automatically generates:
- Ultra-realistic images
- Character animations
- Voice generation/cloning
- Lip-sync
- Background audio
- Final video with a full timeline
- No manual steps are required, but users can review/tweak results afterward.

2. **Smart Director – Manual Mode**
Users write a script, click the "Smart Director – Manual" button, and control each production stage (e.g., generate image, skip voice, add audio) step-by-step.
Generated elements are added to the final timeline — ideal for advanced users seeking precise control.

3. **Standalone Mode**
Users can use individual tools (image generation, voice cloning, animation, lip-sync, background audio) via dedicated buttons for quick tasks or partial content creation (e.g., podcasts, teasers).
Manual assembly or import into the editor is required for final video output.

---

## Project Scope (Milestones and Tasks)

### Milestone 1: Ultra-Realistic Image Generation ($1200-$1400, 2-3 weeks)
- **Tasks**:
- Develop an ultra-realistic image generation system for characters and backgrounds, supporting high-fidelity outputs for video integration.
- Image generation by:
- Text to Image
- Real image to Image (AI)
- Integrate with Character Pack (image + voice) for consistent character identity across scenes.

- **Technical Requirements**:
- Use open-source models like Stable Diffusion or DALL·E-style architectures.
- Optimize for RTX 5090 with CUDA and FP16 for performance.
- Ensure compatibility with animation and video modules.

---

### Milestone 2: Voice Generation & Sound Library ($1000-$1300, 2-3 weeks)
- **Tasks**:
- Develop a voice generation and cloning system supporting 8+ characters with studio-grade audio (48kHz, natural breathing, pauses, emotional tones).
- Support all Arabic dialects (e.g., Egyptian, Gulf, Levantine, Moroccan) with 100% voice cloning accuracy.
- Integrate a local sound library and Pixabay search for background audio (music, applause, etc.).
- Implement a background audio engine for multi-layered audio tracks with seamless blending.
- Add a "Skip This Step" button for voice and sound stages, integrated with Smart Director.

- **Technical Requirements**:
- Use open-source models like VITS or Tacotron 2 for voice generation/cloning.
- Optimize for performance (8-14 seconds per 10-15 word voice line) using CUDA on RTX 5090.
- Integrate with Character Pack from Milestone 1.

---

### Milestone 3: Image Animation & Lip Sync ($1200-$1600, 2-3 weeks)
- **Tasks**:
- Develop an image animation system for full-body motion (eyes, facial expressions, shoulders, hands, legs) and cinematic background motion (e.g., moving trees, people walking).
- Implement lip-sync and full-body motion matching or exceeding DreamFace or SkyReels-level quality, supporting four animation modes: Narration, Dialogue, Mixed, and No Lip Sync.
- Add a "Skip This Step" button for animation, integrated with Smart Director.
- Ensure animations are generated in 12-20 seconds per image with high-fidelity output.

- **Technical Requirements**:
- Use open-source models like SadTalker or FirstOrderMotion for animation.
- Optimize for RTX 5090 with CUDA and FP16 for performance.
- Integrate with Character Pack and voice modules from previous milestones.

---

### Milestone 4: Chat Window (Hasaballa GPT) & Smart Director ($2000-$2300, 2-3 weeks)
- **Tasks**:
- Build a chat window powered by an offline AI model (e.g., Mixtral 8x7B or Nous-Hermes-2 Mistral 7B) supporting all Arabic dialects and script analysis (15 scenes in 6-10 seconds).
- Support text input (15,000 characters), voice commands (speech-to-text), and upload of 8+ reference images/voices.
- Implement intelligent script parsing for scenes, backgrounds, sounds, and lip-sync instructions.
- Develop Smart Director with Auto and Manual modes, including a visual progress bar for Manual mode.
- Add optional internet access buttons ("Smart Internet Access" and "Disconnect") and a saved projects panel.

- **Technical Requirements**:
- Optimize AI model for offline use with CUDA on RTX 5090.
- Ensure full RTL support for the Arabic interface.
- Integrate with previous modules (image, voice, animation).

---

### Milestone 5: Final Video, Editing, and Integrations ($1000-$1400, 3 weeks)
- **Tasks**:
- Develop a CapCut-style video editor with trimming, transitions, subtitles (manual/AI-generated), filters, zoom/pan, and drag-and-drop timeline.
- Support video export in HD, Full HD, 720p, 1080p, 4K with aspect ratios (16:9, 9:16, 1:1).
- Implement Smart Edit Layer for editing via Arabic prompts with timecodes.
- Integrate with Google AdSense and YouTube Studio for revenue tracking and performance analytics (requires internet connection button).
- Build a 100% Arabic RTL interface with large buttons and tooltips.
- Add auto-update, save project, project folder system, and backup/restore functions.
- Fully connect and integrate all modules (image generation, voice, animation, GPT chat, and timeline editor) into a unified, seamless platform** — ensuring smooth transitions and full workflow continuity.
- Provide full source code ownership upon completion.
- **Technical Requirements**:
- Use FFmpeg for video processing and export.
- Implement APIs for AdSense/YouTube integration.
- Optimize for RTX 5090 with CUDA and FP16.
- Ensure secure file management and backup systems.

---

## Required Skills
- **AI & Machine Learning**: Open-source models like Stable Diffusion, VITS, Tacotron 2, SadTalker, Mixtral 8x7B.
- **Video Processing**: FFmpeg, CapCut-style editing tools.
- **Web Development**: HTML, JS, React – with full RTL Arabic interface design.
- **API Integration**: Google AdSense, YouTube Studio.
- **System Design**: Secure file systems, offline-first architecture.
- **Hardware Optimization**: CUDA, FP16, RTX 5090.
- **Arabic Fluency**: Understanding dialects, full Arabic support.

---

## Platform Requirements
- Full offline operation (up to 200GB)
- Image gen: 10–18 sec/image
- Voice gen: 8–14 sec / 10–15 words
- Animation: 12–20 sec/image
- Script analysis: 15 scenes in 6–10 sec
- Video output: 7 mins in 45 mins / 15 mins in 90 mins
- Must match or exceed reference video quality

---

## Budget & Timeline
- **Total Budget**: $6400–$8000
- **Milestone Breakdown**:
- M1: $1200–$1400
- M2: $1000–$1300
- M3: $1200–$1600
- M4: $2000–$2300
- M5: $1000–$1400
- **Timeline**: 11–15 weeks total

*Note*: Milestone 5 may need increased time for quality — 3–4 weeks.

"We aim to maximize quality within our total budget cap of $8000, prioritizing efficient resource allocation across all milestones."
---

## Deliverables
- Fully functional Hasaballa AI platform
- Source code with docs (Python, JS, etc.)
- Optimized for RTX 5090
- RTL Arabic UI
- Full source code ownership

---

## Team Requirements
- Team of 2–6 devs (AI, video, front-end, back-end)
- Senior or mid-level
- Weekly progress updates
- Portfolio required

---

## How to Apply
Please include:
1. Breakdown per milestone
2. Relevant portfolio links
3. Timeline + budget confirmation
4. Team composition
5. Availability to start now
6. Developers must optimize the platform (up to 200GB) for efficient performance on the specified hardware using techniques like FP16 and model compression.

---

Notes
- Full offline operation (Pixabay, AdSense, YouTube optional online)
- No proprietary code unless open-license
- Must review quality links
- Ownership must be fully transferred

Apply now and help build the next-generation Arabic AI video platform!" (client-provided description)