Secrets AI Video Generator: How It Works, Quality, and Cost
Video generation from AI companion images is a genuinely rare capability in the AI girlfriend category. Most major platforms — Character.AI, CrushOn AI, Janitor AI — don't offer it at all. Secrets AI's video generator is not a secondary feature; it's one of the platform's two clearest competitive differentiators alongside its memory system.
This page covers how the generation process works, realistic quality expectations, exact Moments costs by clip length, and the honest budget math for each subscription tier.
What the Video Generator Actually Does
The Secrets AI video generator converts static AI companion images into short animated video clips using a text prompt to direct the movement or action. The output is a rendered video file — not a live interactive stream or a video call — showing the companion character moving, expressing, or responding to the specified scenario.
This is meaningfully different from what most AI image platforms offer. Most competitors stop at static image generation. The video layer adds motion to the companion experience: facial expressions, body movement, contextual gestures. For users where visual content is central to the companion experience, this distinction matters.
The feature is available from the Lite tier ($5.99/mo) upward. Free accounts cannot generate video. The Lite restriction limits output to 3-second short clips; full-length video generation requires Plus ($9.99) or higher.
Most competing platforms do not offer this capability:
- Character.AI (KG: /g/11sck8d802): No video generation
- CrushOn AI: No video generation
- Janitor AI: No video generation
- Candy AI: Limited video generation capability
Secrets AI's video generator is a genuine category differentiator, not a marketing claim.
How to Generate a Video: Step by Step
The generation workflow:
- Start from an existing companion image — either an auto-generated character image or one you've previously generated in the chat
- Select the image as the source for video generation
- Write a text prompt describing the desired movement, action, or expression (specific prompts produce more accurate results than vague ones)
- Submit the generation request — the AI begins processing
- Wait approximately 2 minutes for rendering to complete
- Review and save the completed video clip
The generation time of approximately 2 minutes is consistent across reported user experiences. Unlike image generation (which completes in seconds), video generation involves substantially more computational processing — the wait time reflects that complexity.
Videos are context-aware: the rendering takes into account the character's established appearance and, where relevant, the current conversation scenario. A video generated mid-roleplay can reflect the emotional or situational context of that moment.
Quality: What Reviewers Actually Say
External reviewers at aigirlfriendscout rate the Secrets AI video generator 4.1/5. The consistent characterization from user feedback: videos "look good and move smoothly most of the time."
More specifically, the quality assessment breaks down as:
- Character movement: Natural, not mechanical or jerky in most outputs
- Facial expressions: Realistic in the majority of generations
- Consistency with source image: Character appearance carries through accurately from the static image
- Prompt accuracy: Specific, well-structured prompts produce results closer to the intended action; vague prompts produce less predictable outputs
- Quality variance: Some prompts produce better results than others; occasional outputs don't match expectations
The 4.1/5 rating is strong for the category — it reflects a genuinely functional feature rather than a showcase capability. Quality does improve on Premium generation models, which are accessible on Premium and Ultimate tier subscriptions.
The technology underlying AI video generation (applications of deep learning (KG: /m/0h1fn8h) and video synthesis models related to architectures like Stable Diffusion (KG: /g/11tcd8vgn9)) is advancing rapidly, and the quality floor for consumer-accessible AI video is rising. Secrets AI's implementation sits at a competitive level for its price point.
Moments Costs: The Real Numbers
Video is the most Moments-intensive feature on the platform. Understanding the exact cost structure prevents budget surprises.
| Video Type | Moments Cost |
|---|---|
| Short clip (3 seconds) | ~50 Moments |
| Full-length clip | ~600 Moments |
For comparison against other features:
| Feature | Cost (Moments) | What You Get |
|---|---|---|
| Text message | 1-2 | Single response |
| Image generation | 25-50 | One static image |
| Short video (3s) | ~50 | Brief motion clip |
| Full video | ~600 | Longer motion clip |
| Voice call | 100/minute | Real-time audio |
The trade-off is clear: one full video costs the same as 12-24 images, or 6 minutes of voice, or 300-600 text messages. Video is a premium feature in Moments terms even if it's technically included in your subscription tier.
Budget math by tier:
| Tier | Monthly Moments | Full Videos Possible | Short Videos Possible |
|---|---|---|---|
| Lite | 1,000 | ~1-2 | ~20 |
| Plus | 3,000 | ~5 | ~60 |
| Premium | 8,000 | ~13 | ~160 |
| Ultimate | 15,000 | ~25 | ~300 |
These numbers assume all Moments go to video. In practice, Moments are split across images, text, voice, and memory saves — actual video output will be lower if you use other features too.
Key insight: Heavy video users will likely exhaust their monthly Moments allocation before the billing cycle resets. On Premium, generating even 5-6 long videos per week requires top-up purchases or an Ultimate subscription. The free vs premium comparison shows how the Moments math works across all features simultaneously.
Tips for Better Video Output
The quality of a video generation depends significantly on how you approach the prompt and source image:
Use high-quality source images. The video output is only as good as the input image. Character images generated on Premium or Advanced generation models produce sharper, more detailed video results than those generated on standard models.
Write specific prompts. "Walking slowly toward the camera with a smile" produces a more accurate result than "walking." Describe the movement type, direction, speed, and expression separately rather than in vague combined phrases.
Test with short clips first. A 3-second clip at ~50 Moments lets you verify that your prompt direction works before committing 600 Moments to a full-length version. This is especially important for new types of prompts or actions you haven't tested before.
Contextual prompts work well. If your conversation has established a specific scenario — a particular setting, mood, or action sequence — prompts that reference that context produce outputs more coherent with the overall companion experience.
Save the best source images. If a particular character image generates video well (smooth movement, accurate expression), note what made it work — pose, lighting direction, facial angle — and replicate those factors in future image generation requests.
Who Gets Value from the Video Generator
The video generator is worth the Moments investment for some users and not for others:
Worth using if:
- Visual content is a core part of why you use the platform
- You generate and save companion media regularly
- The companion experience for you includes seeing the character in motion, not just in static images
- You're on Premium or Ultimate where the Moments budget can support regular video use
Not worth prioritizing if:
- Your primary use is text conversation and the visual layer is incidental
- You're on a tight Moments budget where video consumption would crowd out other features
- You're on the free tier (not available) or Lite (limited to 3-second clips)
Best tier for video: Ultimate ($39.99) for consistent heavy video output — 15,000 Moments monthly supports roughly 25 full-length videos. Premium ($19.99) for moderate use — 8,000 Moments supports approximately 13 full-length videos alongside other feature usage.
Competitive Position: Who Else Offers Video?
The AI companion market's competitive map on video generation:
- Candy AI: Limited video — less developed than Secrets AI's implementation
- Character.AI: No video generation
- CrushOn AI: No video generation
- Janitor AI: No video generation
- GirlfriendGPT: No video generation
- Replika: No video generation
- SweetDream AI: Video available (niche competitor)
- Xotic AI: 4K 15-second clips (higher-quality niche option)
The practical comparison: if video generation from your AI companion is a requirement, Secrets AI is currently the most accessible mainstream platform offering it. SweetDream AI and Xotic AI exist in the same space but with smaller user bases and less established track records.
This competitive advantage is Secrets AI's clearest argument for users weighing it against alternatives. The full comparison with competing platforms in the main review contextualizes this against Secrets AI's weaknesses like the no-native-app limitation.
For the detailed Moments pricing including top-up bundle options for video-heavy users, see the pricing page.
FAQ
Short clip duration is approximately 3 seconds, available on Lite tier and above at roughly 50 Moments per clip. Full-length clips are longer animated segments available on Plus tier and above, costing up to 600 Moments per clip. The platform doesn't specify a precise maximum duration for full-length videos in its public documentation, but "longer clips" at 600 Moments is the primary production format for paid subscribers.
No. Video generation is not available on the free tier. A minimum of the Lite subscription ($5.99/mo) is required to access video generation, and Lite is limited to 3-second short clips. Full-length video generation requires Plus ($9.99/mo) or higher, plus sufficient Moments balance to cover the generation cost (up to 600 Moments per full clip).
This depends on your tier's Moments allocation and whether you use Moments for other features. Dedicated entirely to full-length videos (600 Moments each): Lite supports ~1-2 per month, Plus ~5, Premium ~13, and Ultimate ~25. In practice, most users split Moments across images, voice, and video — so actual video output will be lower than the maximum theoretical capacity.
Rated 4.1/5 by external reviewers, the video quality is described as natural character movement and realistic facial expressions in most outputs. The quality is above average for consumer-accessible AI video generation. Some outputs fall short depending on prompt complexity and source image quality. Using high-quality source images and specific directional prompts consistently produces better results. Premium generation model access (Premium and Ultimate tiers) improves quality relative to the standard model.