Secrets AI Video Generator: How It Works, Quality, and Cost
Video generation is the feature that most clearly separates Secrets AI from the competition. The overwhelming majority of AI companion platforms do not offer it — Character.AI, CrushOn AI, and Janitor AI all lack the capability. Candy AI has limited video support. Secrets AI has built video generation into its core feature set and made it accessible from the Lite tier upward.
This guide covers how the generation process works, what the output quality looks like in practice, the full Moments cost structure, and who the feature is actually worth it for.
What the Video Generator Does
Secrets AI's video generator converts a static AI companion image into a short motion clip. The source is an image — either one the platform generated automatically when you created your character, or an image you generated separately in chat. You supply a text prompt describing the movement or action, and the AI processes this into a video clip.
The result is a companion-consistent short video: your character, with their established appearance and visual profile, performing the motion you specified. Because the video derives from an existing image rather than starting from scratch, character appearance remains consistent with what you see in static image generation.
This feature is genuinely rare in the AI companion market. Deep learning-based video generation from a reference image requires significantly more computational resources than static image generation — which explains why most platforms have not implemented it. The platforms that do offer something comparable include SweetDream AI and Xotic AI (which generates 4K 15-second clips), but the mainstream field has not caught up.
For a broader view of how video fits into the platform's full capability set, the complete features overview covers every feature in context.
Step-by-Step: Creating a Video
The generation process has four stages:
- Select or generate a source image. Open a conversation with your companion and either use an existing character image or generate a new one (25-50 Moments). The video will inherit the character's appearance from this image.
- Write a video prompt. Describe the movement or action you want. Specific prompts produce better results than vague ones — "walk toward the camera and smile" yields more predictable output than "do something."
- Submit and wait. Processing takes approximately 2 minutes. The platform is running image-to-video deep learning inference, which is computationally intensive. Do not expect near-instant output like static image generation.
- View and save. The completed clip is delivered in the chat interface. Save it locally if you want to keep it.
The context-awareness of the system means the video reflects the character's established appearance and the scenario context of the conversation. The same prompt applied to two different characters will produce meaningfully different outputs based on each character's visual profile.
Video Quality: Honest Assessment
Independent reviewers rate Secrets AI video quality at 4.1/5. That score aligns with practical testing results — the output is notably better than what most users expect from an AI companion platform, and genuinely impressive by the standards of the current market.
Specific quality characteristics:
- Character movement is smooth in most outputs, with natural transitions rather than jerky frame-to-frame shifts
- Facial expressions animate convincingly — this is one of the technically harder aspects of video generation and Secrets AI handles it better than expected
- Character appearance consistency with the source image is strong — the person in the video looks like the person in the image
- Occasional quality variations appear with complex prompt scenarios (multi-person interactions, unusual movement patterns, backgrounds with many elements)
- The Premium generation model produces noticeably better video output than the standard model — the quality difference is more visible in video than in static images
The known limitation: video generation quality currently does not match the best dedicated AI video platforms (like Runway or Sora for general content). Within the AI companion category, however, it is the current quality standard.
Moments Cost Structure
Understanding the cost before generating avoids budget surprises. Video is the most Moments-intensive feature on the platform:
| Video Length | Moments Cost |
|---|---|
| Short clip (3 seconds) | ~50 Moments |
| Standard clip | ~100-300 Moments |
| Full-length clip | ~600 Moments |
For comparison, the same Moments spent differently:
| 600 Moments equals... | Quantity |
|---|---|
| 1 full video clip | 1 clip |
| Images at 50 Moments | 12 images |
| Voice call at 100/min | 6 minutes |
| Short video clips at 50 | 12 short clips |
The cost comparison illustrates why video budget planning matters. A single full-length video clip consumes as many Moments as 12 images or 6 minutes of voice calls.
Monthly Video Budget by Subscription Tier
| Tier | Monthly Moments | Short Videos (~50 ea) | Full Videos (~600 ea) |
|---|---|---|---|
| Lite | 1,000 | ~20 clips | ~1-2 clips |
| Plus | 3,000 | ~60 clips | ~5 clips |
| Premium | 8,000 | ~160 clips | ~13 clips |
| Ultimate | 15,000 | ~300 clips | ~25 clips |
These figures assume all Moments are spent on video — realistic mixed usage (combining text, images, and video) will reduce these numbers. For practical planning:
- Lite ($5.99/month): 1,000 Moments supports light video exploration. One or two short clips plus some images is realistic.
- Plus ($9.99/month): 3,000 Moments supports regular video use alongside images. 3-5 full clips per month with room for other usage.
- Premium ($19.99/month): 8,000 Moments is the comfortable tier for users who generate video regularly without going to extremes.
- Ultimate ($39.99/month): 15,000 Moments for heavy video creators who want volume without constant top-up purchases.
If you anticipate heavy video use, buying additional Moments from the top-up store is available independently of subscription tier: bundles start at 1,980 Moments for $5.99, scaling up to 118,800 Moments for $249.99. Premium and Ultimate subscribers receive 10-15% bonus on top-up purchases.
For the complete pricing breakdown across all tiers and Moments costs, the pricing and Moments guide covers every number. For a tier-by-tier capability comparison including which tiers unlock video, the free vs premium breakdown is the direct reference.
Practical Tips for Better Video Output
From testing video generation extensively, these practices improve results:
- Start with short clips before long ones. A 50-Moments short clip gives you a quality preview before committing 600 Moments to a full clip. If the motion looks wrong on the short version, adjust the prompt before scaling up.
- Use high-quality source images. The video inherits whatever artifacts or quality issues exist in the source image. Generate a clean, high-quality image specifically for video conversion rather than using a lower-quality character photo.
- Keep prompts action-specific. "Smiling warmly and tilting her head" generates more predictable output than "be friendly." Concrete, single-motion prompts outperform vague emotional descriptors.
- Use the Premium generation model. The quality difference between standard and Premium generation models is more visible in video than in static images. If you are spending 600 Moments on a full clip, use the best model available on your tier.
- Generate at off-peak hours. Peak traffic periods slow down video generation. The 2-minute typical processing time can extend significantly during high-demand periods. Early morning and midday generally produce faster results than evenings.
- Save Moments by sequencing correctly. Generate an image, evaluate its quality, then convert the best one to video rather than generating video directly from lower-quality auto-generated character images.
Who Should Use the Video Generator
Worth the Moments investment if:
- Visual content alongside conversation is an important part of your companion experience
- You want unique, personalized media from your specific character rather than generic AI videos
- You are on Premium or Ultimate and have the Moments budget without depleting other usage
- You enjoy creating and saving companion content as a form of engagement
Better to skip or minimize if:
- You are on the free tier (unavailable) or Lite and want to stretch Moments across the month
- Your primary engagement is text-based conversation — the Moments are better spent on memory features or images
- Your Moments budget is tight and video would crowd out other usage types
- You are still evaluating the platform and haven't committed to extended use
The Competitive Position
No direct survey of every AI companion platform has found video generation as a standard feature. The platforms that are most directly relevant:
- Character.AI: No video generation
- CrushOn AI: No video generation
- Janitor AI: No video generation
- Candy AI: Limited video (less developed implementation)
- Replika: No video generation
- GirlfriendGPT: No video generation
The rarity of the feature makes it genuinely distinctive rather than a checkbox capability. For the full review covering how video fits into the overall platform assessment, see the complete Secrets AI review.
Access the Video Generator on Secrets AI →
FAQ
Video length varies by generation settings and Moments budget. Short clips are approximately 3 seconds and cost around 50 Moments. Full-length clips are longer and can cost up to 600 Moments per clip. The Lite tier supports 3-second clips; higher tiers unlock access to longer generation options.
No. Video generation is not available on the free tier. It requires at minimum a Lite subscription ($5.99/month). Free accounts also cannot generate images — media generation of any kind requires a paid plan. The 200 one-time Moments on the free tier are available only for text-adjacent features.
The number depends on your subscription tier and clip length. On Plus (3,000 Moments), expect approximately 5 full-length videos or 60 short 3-second clips before exhausting the allocation — assuming you spend all Moments on video. Mixed usage (combining images, voice, and text) reduces video count accordingly. On Premium (8,000 Moments), approximately 13 full clips or 160 short clips are possible per month in a video-only scenario.
Video quality is rated 4.1/5 by independent reviewers. Character movement and facial expressions render realistically in most outputs. Appearance consistency with the source image is strong. Complex prompts or unusual movement patterns occasionally produce lower-quality outputs. Using the Premium generation model and high-quality source images produces the best results. The quality is competitive within the AI companion category, though it does not match specialized AI video production platforms designed for general-purpose video creation.