AI images, videos and avatars

Repost21 can create media for your posts right inside the composer: images from a text description, carousels for Instagram and LinkedIn, short videos and talking heads with voiceover. Everything is generated in the same window where you write your post and is attached to it instantly. And the built-in editor lets you polish the image — add text, replace the background, crop.

The generation buttons live in the post editor (“Content” → “New post”): Image, Video, Carousel, AI avatar. The same tools are available from the media picker — everywhere files get attached.

Images

Describe in words what should be in the image — and get a ready picture in a few seconds.

1. Click “Image” in the post editor.
2. Write a description: what is shown, in what setting, in what mood.
3. If you like, pick a model, an aspect ratio, a style and a brand library.
4. Click “Generate” — the ready image can be attached to the post in one click, or you can generate another.

Model. You can choose from several models of the Gemini family (Nano Banana) — from fast and cheap to the highest quality. If the top model is overloaded, Repost21 automatically switches to the next one — the generation won’t fail.

Aspect ratio: 1:1, 4:5, 9:16, 16:9 or 3:4 — to match the format of the platform the post is going to.

1 or 3 variants. The “1 / 3” toggle generates three image variants at once in parallel — you pick the best one and attach it.

”Per platform” — the right format for each social network

If the post goes to several social networks at once, the “Per platform” action will generate a separate image for each — in the aspect ratio that looks best in the feed:

Platforms	Format
Instagram, LinkedIn	1:1
Telegram, Threads, Facebook, Reddit	4:5
X	16:9
YouTube, TikTok	9:16

Style presets

Below the description field there’s a row of chips with ready-made styles. A click adds the style to the prompt (clicking again removes it): Cinematic · UGC · Product · Flat-lay · Minimalism · Vivid · Cozy · 3D · Neon · Portrait. You don’t need to be a designer or know the “magic words” — one click sets the mood for the whole image.

Brand library as a reference

Connect a brand library in the generation window — and the AI will draw on your real materials: logos, brand style, product photos, backgrounds. You can choose which asset categories to use as references. Along with the images, the brand’s style description and signature palette are mixed into the generation — the result comes out “in your style” rather than abstract.

Tip

If an autopilot agent has a brand library set, it can generate branded images for posts on its own — see Autopilot.

Carousel

The “Carousel” button generates a series of square slides — 3 or 5 of them — for Instagram and LinkedIn carousels. All slides are created in a single run and attached to the post all at once. For a unified visual style across slides, we recommend specifying a brand library.

Video (Veo)

Generation of short videos from a text description via Google Veo.

• Duration: 4, 6 or 8 seconds.
• Format: horizontal 16:9 or vertical 9:16, resolution 720p or 1080p.
• Models: fast (default), basic for less, and premium with maximum quality.
• First frame from the brand library: you can set a reference image — the video will start with it.

Camerawork presets

The chips under the description add ready-made cinematic techniques to the prompt: Crash Zoom · Bullet Time · 360° · Dolly in · FPV · Slow-mo · Hyperlapse · Panorama · Crane · Trailer. One click — and the clip looks like it was shot by a cinematographer, not a “neural-net splash screen.”

Generation in the background

A video renders from 30 seconds to a couple of minutes. You don’t have to wait in the modal: the “Minimize to background” button hides the window while the task keeps running — a card with progress appears in the tray at the bottom right. You can switch to another page and get back to work; when the video is ready, you’ll attach it from the tray.

Note

Videos are generated without sound — that’s a limitation of the current Veo integration. Add music or voiceover at the editing stage, or use an AI avatar if you need a talking character.

AI avatar (HeyGen)

A “talking head”: you write a script — the avatar says it with lip movement and facial expressions. 175+ languages are supported, including Russian and Uzbek.

1. Click “AI avatar” in the post editor.
2. Write a script — the text the avatar will say (up to 5000 characters). Next to it you’ll see an estimate of duration and cost.
3. Pick an avatar from the catalog (a grid with previews and search) and a voice (with a language filter and sample playback).
4. Set the format — 16:9, 9:16 or 1:1 — and, if you like, a background color.
5. Start the generation. The clip takes 1–5 minutes; like a video, the task can be minimized to the background tray so you can keep working.

Your own digital twin from a photo

You can speak “with your own face”: upload a portrait photo to the brand library in the “Person” category and click “Make AI avatar” on the card. After that the photo appears in the “My avatars” section in the generation window — you pick it instead of a catalog avatar, and the clip is voiced with your face.

Note

The photo must show a clearly distinguishable face — otherwise avatar creation won’t pass the check.

Image editor

Any image — generated or uploaded — can be opened in the built-in editor (the edit button on the result preview or on the card in the media library).

Capability	What it does
Layers	Text, rectangles, circles, lines, images on top of the original; move, scale, rotate, transparency
Filters	Brightness, contrast, saturation, blur + ready-made presets (Vintage, B/W, Warm, Cool, Vivid, Soft)
Cropping	Free or guided by 1:1 / 4:5 / 9:16 / 16:9 / 3:4
AI edit	Describe the change in words (“replace the background with a sunset”) — the result lands as a new layer or replaces the background
Brush mask	Paint an area with the brush and describe what to change in it — the AI repaints only the selection
Versions	Every save is a new version with history; you can revert to any one or overwrite the original
Export	PNG, JPG or WebP, with quality and 1×/2×/3× scale settings

The usual hotkeys work: Cmd/Ctrl+Z — undo, precise layer nudging with arrow keys, double-click on text — edit in place.

Cost and limits

• An estimated cost is shown before generation — right in the window, together with your remaining monthly quota. For video the price depends on the model and duration, for the avatar — on the length of the script.
• Availability by plan: image generation — from STARTER, video and AI avatars — from PRO. Monthly quotas (number of images, seconds of video and avatars) depend on the plan — see the current values in the “Plan” section in the app or on the Pricing page.
• The quota is charged only for successful generations.