Secrets AI Video Generator: How It Works, Quality, and Cost

Video generation from AI companion images is the single feature that most clearly separates Secrets AI from the competition. Character.AI does not offer it. CrushOn AI does not offer it. Janitor AI does not offer it. Candy AI has limited video capability but nothing comparable in scope. If you want to turn your AI companion's static image into a short motion clip, the available mainstream options narrow to a short list — and Secrets AI is at the top of it.

This page covers the mechanics of the video generator, honest quality assessment, the full Moments cost math, and which tier makes video use financially sustainable.

For the full platform assessment, see the full review.

What the Video Generator Actually Is

The video generator converts a static AI companion image into a short animated clip based on a text prompt. The output is a brief video featuring your companion with motion — movement, expressions, and animation derived from the source image and guided by your written description.

This is distinct from video chat or real-time interactive video — there is no live rendering or two-way video interaction. The workflow is closer to AI animation: provide an input image, describe the desired action, receive a rendered clip approximately two minutes later.

The feature is available on Lite tier and above. Free accounts cannot access video generation regardless of Moments balance. Most competitors do not offer this feature at all — which makes it genuinely distinctive in the AI companion market, not just a marketing differentiator.

How Video Generation Works: Step by Step

The process from start to delivered clip:

Step 1: Generate or select a source image. Either generate a new companion image (25-50 Moments) or select an existing one from your character's image library. Image quality directly influences video output quality — higher-quality source images generally produce better-rendered video clips.

Step 2: Add a text prompt. Write a description of the desired motion, action, or scenario. The prompt guides what the character does in the clip. Specific, concrete prompts ("walking slowly across the room, turning to look at the camera") produce more reliable results than abstract descriptions.

Step 3: Select generation quality. Higher tier subscriptions access the premium generation model, which produces better motion quality. On Premium and Ultimate, the Advanced model option is available.

Step 4: Wait for rendering. Processing takes approximately two minutes for a standard clip. This is not instant — plan for the generation time in your usage pattern.

Step 5: View and save. The rendered clip appears in the chat interface. Save it directly if you want to keep it.

Video Quality Assessment

Reviewers rate the video generator 4.1/5 — described as "look good and move smoothly most of the time." That qualifier matters: most of the time, not always. Quality varies based on prompt complexity, source image quality, and the generation model used.

What typically works well:

• Simple, single-action movements (walking, turning, looking toward camera)
• Facial expressions and upper-body motion
• Character consistency with the source image
• Natural fluid motion on straightforward prompts

Where quality sometimes falters:

• Complex multi-step action sequences in a single short clip
• Hand and finger rendering (consistent with broader AI generation limitations)
• Very rapid motion or physically complex poses
• Long clips attempting detailed environmental context

The 4.1/5 rating is honest — it reflects a feature that delivers on its core promise without being exceptional. For a platform where video generation is a genuine rarity, setting expectations at "solid, not perfect" is the accurate framing.

Moments Cost — The Complete Math

Video is the most Moments-intensive feature on the platform. Understanding the cost structure prevents the most common mistake: upgrading to Plus, generating video freely, and running out of Moments two weeks into the billing cycle.

Budget per tier for video use:

Cross-feature cost comparison (600 Moments):

For the same Moments cost as one full video clip, you could alternatively generate 12-24 images, or have 6 minutes of voice calls, or send several hundred text messages. Video is expensive relative to every other media type. This is not hidden — it is the platform's documented cost structure.

The practical implication: If you want to use video generation regularly (multiple long clips per week), Premium ($19.99/month) is the minimum viable tier. Plus ($9.99/month) allows approximately 5 long clips per month — one per week, approximately, with Moments to spare for images and voice.

Video vs Images vs Voice — Putting Costs in Context

When making Moments allocation decisions, this comparison helps:

Text is essentially free in Moments terms — you can send 3,000 messages on a Plus account for the same cost as 6-12 images. Images are moderate cost. Voice is expensive per minute but predictable. Video (long clips) is by far the most Moments-intensive choice per unit of content.

For the Moments costs by tier and bulk purchase options detail, the pricing page has the full breakdown including top-up bundle pricing.

Tips for Getting Better Video Results

Practical guidance based on how the generation system works:

Invest in quality source images first. The video renderer uses the source image as its reference. A high-quality, well-lit, properly composed image in your desired setting produces better video than a lower-quality source image. Spend Moments on a good image before converting to video.

Write specific motion prompts. "Smiling and nodding" is more reliable than "being happy." Concrete physical actions produce more consistent results than emotional states.

Start short to test quality. Generate a 3-second clip (~50 Moments) with a new prompt before committing to a full 600-Moment long clip. If the short clip quality meets expectations, the longer format is worth the investment.

Use the premium generation model on Premium/Ultimate. The visual quality difference between standard and advanced generation models is noticeable on video more than on static images.

Match scene context to your prompt. If your character is in a specific setting or outfit in the source image, your motion prompt should be consistent with that context. Mismatches between the source image and the prompt description reduce output coherence.

Who Should Use the Video Generator?

Video generation makes sense if:

• Visual content is a significant part of your AI companion experience
• You want personalized dynamic media, not just static images
• You are on Plus or higher tier and can budget 5+ clips per month within your Moments allocation
• You value Secrets AI's specific differentiator over competitors that cost less but lack this feature

Video generation is not worth the Moments if:

• Text-based conversation is your primary use and visual media is secondary
• You are on the Lite plan and want to preserve Moments for regular image generation
• You are budget-constrained — at 600 Moments per long clip, the cost adds up quickly
• You prefer spending Moments on the volume of images over fewer video clips

Competitors with Video Generation

This feature's rarity is documented and genuine. The competitive landscape for AI companion video generation:

• Character.AI (KG: /g/11sck8d802): No video generation
• CrushOn AI: No video generation
• Janitor AI (KG: /g/11njfp42__): No video generation
• Candy AI: Limited video capability; less developed than Secrets AI's implementation
• GirlfriendGPT: No video generation
• SweetDream AI: Comparable video capability (limited mainstream presence)
• Xotic AI: Offers 4K 15-second clips (higher quality ceiling, more niche platform)

For the AI companion market specifically — where deep learning-based image generation via tools related to Stable Diffusion (KG: /g/11tcd8vgn9) has become standard — video generation remains technically differentiated. The computational overhead and generation time (approximately 2 minutes per clip) indicate why most platforms have not implemented it.

For video access by tier and all platform features beyond video, those pages complete the picture.

FAQ