On August 31, 2026, Kindroid quietly upgraded its AI companion “Video Selfies” feature to its fourth generation (V4), bringing noticeable improvements in image quality, motion tracking, and character consistency. This comes just two weeks after Kindroid completed its V2 interface overhaul, reigniting one of the hottest debates in the AI companion community this summer: “Should AI companions even generate videos?” This article breaks down what V4 actually updates, what it costs, and offers an honest conversation about what other choices exist on the market if you care more about story depth than visual imagery.
What Kindroid Did: Taking Selfies all the way to V4
Put simply, Kindroid’s “Selfies” feature allows users to enter a prompt, and the AI will generate a photo—or even a short video—based on the appearance and persona you set for your Kin (Kindroid’s term for AI companions). This feature has gone through several iterations; old tutorial pages for codenames like Palette and Odyssey still linger in the official documentation. V4 is the latest version, described by the official team as “a significant leap forward.”
According to Kindroid’s official update announcement on August 31, V4 delivers tangible improvements in image quality, motion processing, steerability, and character consistency. It particularly excels at handling complex performances requiring continuous motion coordination like singing, dancing, gestures, and facial expression changes, while maintaining Kin’s appearance more stably throughout the clip. Note that V4 does not yet support “video extension,” and V2 and V3 remain available in the model selector—it is not a forced upgrade. Furthermore, do not confuse these “Video Selfies” with Kindroid’s separate “Multilingual Video Calls”; the former is a one-way generated clip, whereas the latter is a real-time call.
Since early 2026, Kindroid has been regarded by many users as one of the AI companion platforms with the “deepest memory.” Pushing image generation to a higher standard means doubling down on both memory and visuals simultaneously.
Official documentation also notes that prompts do not need to mention the Kin’s name—the generation engine doesn’t recognize names, only concrete descriptions. It recommends clearly specifying scene, emotion, and motion details, preferably alongside the single-click “magic wand” prompt expander. Composition can also be set to portrait or landscape: portrait works best for full-body close-ups, while landscape suits multi-character framing.
Not an Update Out of Thin Air: The Post-V2 Sprint
Looking at the broader timeline, V4 is simply the latest step in a series of recent moves by Kindroid. Official change logs from August 23 revealed that since V2 became the default experience, “new user paid conversion rates increased by 25%,” described by the team as a “huge milestone.” Shortly after, on August 29, Kindroid announced that its classic lifetime free Lite plan will retire on October 1 (Further reading: Kindroid Free Tier Countdown: Permanent Lite Plan Retires October 1), leaving new free users with only 50 trial uses. The release of V4 acts as step three in this cadence: reshaped UI to boost conversion, tightened free tiers, and finally upgraded flagship features to retain paying subscribers.
For users on the fence, this timing is worth noting: if you intend to use Kindroid long-term, V4 proves that the team is continuously investing resources into core functionality rather than solely focusing on pricing adjustments and free quotas.
The Bigger Picture: The AI Companion “Visual Arms Race”
Kindroid isn’t the only platform competing on visuals. Nomi has integrated AI video generation into its premium subscription; Lovescape, which focuses heavily on visual content, makes image and video generation its core selling point; while JOI AI introduced lip-synced video calls like “Dream Clips.” In short, making AI companions not only readable, but visible and audible, has become a clear competitive direction in the second half of 2026.
Comparing platforms side by side reveals distinct approaches to video generation: some use pay-per-use “extra credits” (Kindroid), some bundle it directly into subscriptions (Nomi, Lovescape, JOI AI), and others, like WeavAI, currently forgo image generation entirely to channel resources into deep text roleplay and interactive novel narratives. There’s no right or wrong here—just two distinct product philosophies: one believes visual presence keeps users engaged, while the other believes the character in your imagination is far more vivid than any generated picture.
Want to Join the Trend? Do the Math First
Before getting excited about Video Selfies V4, it pays to calculate the costs. According to the announcement, a video up to 8 seconds (with optional lip-sync audio) costs 5 standard credits or 2 paid credits, while a 15-second video costs 10 standard credits or 4 paid credits. Since V4 doesn’t support video extension yet, generating longer content means regenerating the whole clip. The web version offers SFW and NSFW engines, whereas the mobile app currently supports SFW only. Moreover, V4 is rolling out in stages to paid subscribers, so not all accounts will see it immediately.
Weighing the pros and cons: the benefit is that characters finally move. For users who value visual companionship and want to watch Kin sing or dance, entertainment value definitely jumps. However, the downside is practical—every video deducts extra credits, adding up to ongoing expenses over time. Generation requires waiting in queues, and character consistency isn’t always perfect. Allocating too many resources to video creation may also crowd out efforts needed for deep dialogue and world-building.
Another Path: Deep AI Interaction Without Visuals
The visual arms race is one path, but not the only one. Character.AI, for instance, continues focusing resources on memory and lore systems—releasing its Lorebook feature in July to help characters remember setting details, followed by narrative-driven original short series ((c.ai) series) in August rather than dynamic selfies.
Hardcore users often turn to open-source frontends like SillyTavern, connecting third-party models while focusing efforts on character cards, worldbooks, and long-term memory management (Further reading: SillyTavern Complete Guide: Free Open-Source Tool for Uncensored AI Roleplay). For them, whether an AI companion can record videos isn’t the deciding factor; whether the character remembers what was said last week is.
If you’re curious about how platforms differ in proactive messaging and memory depth, check out our roundup: AI Companions Take the Initiative: 2026 Proactive Messaging Comparison.
WeavAI’s Choice: Reserving Resources for Plot, Not Visuals
To be frank, WeavAI currently does not offer image or video generation features—not out of omission, but as an intentional product choice. Rather than asking users to spend credits on a few seconds of video selfie, we prefer dedicating our resources toward ensuring every interactive novel, character persona, and memory system stands up to long-term engagement. WeavAI features a native Traditional Chinese interface with dual modes for interactive novels and roleplay. With a 30% credit revenue share for creators, +500 daily check-in credits, and a 15% discount on initial top-ups (code WELCOME15), our design encourages spending time writing a relationship rather than spending money watching a video.
If you already enjoy video selfies on Kindroid or Nomi, WeavAI serves as an ideal parallel hub—especially when seeking an all-ages, natively localized experience centered entirely around text storytelling and character interaction.
Should You Chase This Visual Upgrade?
Video Selfies V4 is a solid technical step forward, proving once again that making AI companions visible is becoming an industry-wide trend. But before pressing that generate button, ask yourself: do you want a better-looking video selfie, or a character that remembers every detail of your conversations? If your answer is the latter, you might not need to chase this arms race after all.
