返回资讯中心
外部精选
软件工程
#开发者实时流

Show HN: I built an AI tool that turns two photos into a dance video

Real A2O LITE NEON dance results from this generator — two photos, fixed duo template.

Hacker News 最新fourfire2026/10/20 · 26 天前20 分钟阅读

以下正文同步自 Hacker News 最新,版权归原站所有,已转换为易读排版。

Real A2O LITE NEON dance results from this generator — two photos, fixed duo template.

The A2O LITE AI dance video generator runs in three steps, and none of them require editing experience.

One for each role in the duo routine. Photo 1 maps to the foreground dancer with white sunglasses in the opening shot; Photo 2 maps to the dancer standing behind. Straight-on shots work best — a clear face, even light, no heavy filters.

Every video generated here uses the same two-person NEON stage template: the same framing, the same lighting, the same two-role choreography. The routine never changes — only the two faces do. That is deliberate. A fixed template is what makes the output look consistent instead of random, and it is why a video takes minutes rather than hours.

Preview the result first, then render the full video and download it. Post it wherever you like. If you publish it with the track, credit A2O LITE as the original artist.

Search this term and you will find short-form clips, news pieces and fan compilations rather than a product page. So before the tool, here is the plain answer.

A2O LITE is a girl group under A2O Entertainment, the label founded by Lee Soo-man. The group debuted on October 13, 2026 with the digital single "NEON," performing as a nine-member act. That is the act the term refers to.

A2O's AI angle is Blooming Talk, a fan-interaction service built on AI avatars of the members. Through it, fans can hold voice and video conversations with an AI representation of a member — a companion layer sitting on top of the group, not the group itself.

To be clear about what it is not: the members are real people, not AI-generated characters. The songs are sung by real voices, not synthesised ones. "A2O LITE AI" does not describe a virtual idol group, and it does not describe a voice-cloning product. It describes how one label added an AI interface to a real, human act.

For the longer background, read What Is A2O LITE AI? — a dedicated page for the search phrase, separate from the generator.

The term gets used for three different kinds of clip.

Edited montages that take the debut stage apart and rebuild it for laughs.

Clips where the track is re-sung by a synthesised voice, sometimes in another artist's tone.

Face-swap jobs that drop other people into the choreography.

We do not make any of those three, and that is on purpose. The template here is fixed, the faces are your own or used with permission, and no member's likeness or voice is reproduced. You get the stage — not the stars. If what you came for was one of the three above, this tool will not do it, and no page should pretend otherwise.

The NEON duo template is what most people come here for right now. It is not the only routine the generator handles.

K-pop routines are built on sharp, repeatable point moves, which makes them a good fit for a template-based generator: the camera stays wide, the two roles stay readable, and the identity of the routine survives the swap. If you have ever wanted to appear in a polished K-pop stage without learning the choreography, this is the closest shortcut.

Fast-cut challenge formats work the same way. Pick the routine, drop in two photos, and the cut stays identical from one generation to the next — which is exactly what a challenge format needs.

Short-form dances live and die on recognisability. The template keeps the framing and the beat hits consistent, so the output still reads as that dance even when the faces change.

More routines are added over time. If a format you want is missing, the pricing page shows what is currently available.

"NEON" runs on contrast. The stage is dark, the lighting is saturated, and the choreography leans on mirrored, wide two-person positions rather than a dense centre formation. For a two-role template that is close to ideal: both figures stay visible and equally lit through the whole cut, so neither photo gets pushed into the background.

It also means the template does not depend on facial close-ups. If your photo is slightly soft or shot on a phone, the routine still holds up — the wide framing forgives what a close-up would expose.

The template needs two photos. It does not care who is in them.

The most common use. Two friends, two roles, one video to post on both accounts.

Pets work better than most people expect — a clear, well-lit head shot is enough. Pair yourself with your dog or cat and the foreground role becomes a co-star.

Two separate portraits work fine. They do not need to be taken on the same day, in the same place, or against the same background — the template normalises framing on its side.

If a generation comes back looking off, a sharper photo solves it more often than a different routine does.

The tool is built for your own photos, or photos you have explicit permission to use. That is a hard rule, not a suggestion — it is also what keeps the output clean and the video yours to post.

The template never uses a real person's likeness or voice, including the members of A2O LITE. You are not placing a celebrity into a dance routine; you are placing two photos into a fixed stage. Please do not upload a photo you do not have the right to use.

It is a catch-all phrase that grew out of the group's debut. In practice it points at the wave of AI-labelled edits of the "NEON" stage that spread across short-form platforms in the days after release — parody cuts, synthesised voice covers, face swaps. The members themselves are real performers, and the "AI" in A2O refers to a separate fan-interaction service, not to the group.

No. A2O LITE is a real girl group under A2O Entertainment, and every member is a human performer. The confusion comes from the label's AI side project, which builds AI avatars of the members for fan conversations. An AI feature attached to a real group is not the same thing as an AI group, and only one of those describes A2O LITE.

It is a fan-interaction layer. Fans can hold voice and video conversations with an AI representation of a member — closer to a companion service than to music production. None of it replaces the members' voices on the records, and none of it generates the choreography.

No. This site is independent and has no affiliation with A2O Entertainment, the group, or its members.

No. The template does not accept a real person's likeness, and no member's face or voice is reproduced anywhere in the workflow. Use your own photos, or photos you have clear permission to use.

A preview comes back quickly. A full render takes longer and costs more credits, and the credit cost is shown before you confirm — so you can judge the preview first and only pay for the full video if the framing works.

One clear, well-lit face per photo, shot straight on, with even front lighting and no heavy beauty filters. Phone photos work fine — sharpness matters more than camera quality, and a slightly soft close-up fares worse than a clean wide shot.

Yes. Any two photos work, and pet pairings are one of the more popular uses of the template — a well-lit head shot of a dog or cat is enough to fill the rear role.

Exactly two. Photo 1 maps to the foreground dancer with white sunglasses in the opening; Photo 2 maps to the dancer standing behind.

If you publish the video with the original track, credit A2O LITE as the artist. The tool itself is unaffiliated with them, so there is nothing to credit on our side.

That is handled on our side. Different dimensions, different backgrounds and different lighting between the two photos are all normal — the template normalises framing before the routine is applied.

Yes, and that is the point of the routine list above. The tool is not tied to one release. The NEON duo template is what most people arrive for right now, but the same two-photo flow covers K-pop choreography, random dance challenges and short-form viral routines — so the page stays useful once the trend moves on.

正文由 探How 从来源站点 RSS 同步,内容未经改写;遇到排版缺失或需要图片、视频时请以原文为准。

打开原文
去社区讨论 →