Auto combine
your videos

Clicking Combine Now will pick one random intro, one random main video, and optionally one end video, join them together, and save the result as a direct download.

Upload one mp4. Then on the PC run master.bat command for tt-broadcast (or python -m ig_poster.main tt-broadcast --all) and every phone with TT enabled will, on its next TT turn (~5-10 min), repurpose this video into a unique variant and post to TikTok. Each phone gets a different randomised filter pass so the metadata fingerprint differs per upload. The slot is single-video — uploading again replaces the previous one.

Same flow as TT broadcast but the result is posted as an IG reel. On the PC run master.bat option a (or python -m ig_poster.main ig-broadcast --phones phone-02,phone-03,phone-04) to pick which phones post. Each selected phone repurposes this video into a unique variant and uploads on its next IG turn (or directly in parallel via --now). Daily IG caps + post-gap are bypassed.

Upload one mp4 for IG STORY broadcasting. On the PC run master.bat → sb (or python -m ig_poster.main story-broadcast --video --all) to pick which phones post. Each selected phone repurposes this video into a unique variant and uploads as an IG Story on its next IG turn (or directly in parallel via --now). Independent slot from the IG-reel broadcast — video story + reel can coexist.

Upload one photo (.jpg / .jpeg / .png / .webp) for IG STORY broadcasting. On the PC run master.bat → sb (or python -m ig_poster.main story-broadcast --photo --all). Each selected phone fetches a uniquely-repurposed 1080×1920 jpg from the backend and posts it as an IG Story. Separate slot from the gallery-push photo broadcast — this one is repurposed per phone, that one is same-bytes.

Upload one photo (.jpg / .jpeg / .png / .webp). Then on the PC run master.bat → pp (or python -m ig_poster.main push-photo-all) and every selected phone gets the exact same photo pushed into DCIM/Camera via adb push, with a media-scanner broadcast so the gallery picks it up instantly. No repurpose — same bytes on every phone. The slot is single-photo; re-uploading replaces.

Upload multiple photos (.jpg / .jpeg / .png / .webp / .heic). Each uploaded photo is stored on the backend until you clear it. Then on the PC run master.bat → pm (or python -m ig_poster.main push-gallery-all) and every selected phone gets every photo pushed into DCIM/Camera, with MediaScanner triggered per file so they show up in the phone gallery instantly. By default each phone receives a UNIQUE repurposed copy of every photo (randomized brightness/contrast/saturation/hue/temperature/sharpness/mirror + metadata strip + fresh JPEG encode → different MD5 per phone). Aspect ratio is preserved (square / portrait / landscape all work). Use case: stage a curated stock you'll manually pick from inside each app (story, post, DM, Snap, etc.). Use the master.bat prompt to disable repurpose if you want raw bytes (brand assets, reference shots). Re-pushing later doesn't collide — each push run uses unique filenames.

Upload a single video and pick how many repurposed copies you want (1–10). Each variant runs the standard randomised filter chain independently — so each one ships with a different metadata fingerprint and a slightly different visual treatment. No watermark applied here; this is the manual reposting tool.

Clicking Repurpose Now picks one random video from the folder and re-encodes it with 9 randomised edits — zoom, brightness, contrast, saturation, hue shift, speed, noise, color temperature, and sharpness — so it looks like a brand-new upload.

Same repurpose algorithm as the main one, but pulls from a separate video library. Picks one random video → applies 10 randomised edits → fresh-looking reel.

Same repurpose algorithm (10 randomised edits) over the ZoeReels library. Picks one random video → metadata scramble → fresh-looking reel. No watermark.

Same repurpose algorithm (10 randomised edits) over the AdrianaSillyReels library. Picks one random video → metadata scramble → fresh-looking reel. No watermark.

Same repurpose algorithm (10 randomised edits) over the MiaReels library (new content type). Picks one random video → metadata scramble → fresh-looking reel. No watermark. In the pool but not yet attached to any phone.

Same repurpose algorithm (10 randomised edits) over the BabyReels library (new content type). Picks one random video → metadata scramble → fresh-looking reel. No watermark. In the pool but not yet attached to any phone.

Same repurpose algorithm (10 randomised edits) over the BabyReelsAi library (AI-generated variant of BabyReels, own folder). Picks one random video → metadata scramble → fresh-looking reel. No watermark. In the pool but not yet attached to any phone.

Same repurpose algorithm (10 randomised edits) over the LipSing library (lip-sync content). Picks one random video → metadata scramble → fresh-looking reel. Keeps its own audio. No watermark. In the pool but not yet attached to any phone.

9:16 reel cover images. Each serve keeps the cover's original dimensions — no zoom, no crop, no mirror, so the composed art and text survive intact; only the pixel-level mods (brightness, contrast, saturation, hue, temperature, sharpness) plus a metadata strip and a fresh encode run, which is enough for a unique hash every time. Uncheck the box to download the untouched original. Built for the iPhone farm, which pulls straight from Railway via POST /api/repurpose-nvrsettlecover.

Same repurpose algorithm (10 randomised edits) over the Acrobatic library. Picks one random video → metadata scramble → fresh-looking reel. Keeps its own audio. No watermark. In the pool but not yet attached to any phone.

Picks one random 4-second intro + one random ending, splices them, runs the repurpose algorithm (speed pinned at 1.0x) over the combined video, then overlays a random audio from the third folder — the audio REPLACES the clips' own sound, and the video stops when the audio or the video ends, whichever comes first. "Pick a Video…" selects the intro; ending + audio stay random. No watermark.

V2 is the identical build plus a transparent UI overlay: the 4.05s intro plays clean, then the main video runs at full 9:16 with just the text and icons of a random app frame laid over it — nothing is cropped or squeezed. The overlays come from the prepared batch in the Transparency 🪄 section, so build that batch once before using V2. Both versions are separate posting sources for the phones.

Takes one random clip out of each of the three folders and splices them together in a random order — 1·2·3, 3·2·1, 2·3·1, any of the six. Then the repurpose algorithm (speed pinned at 1.0x) runs once over the whole join, so the three parts read as a single video, and a random sound is laid over the top.

The sound starts 30 seconds into the track and runs until the footage ends, so the output always matches the full length of the three clips together — the track is trimmed there, and the video is never cut short. "Pick a Video…" pins the folder-1 clip; folders 2 and 3, the order and the audio all stay random. No watermark.

Takes one random clip out of each of the four folders (parts 1-4) and splices them together in a random order — any of the 24 possible arrangements. Then the repurpose algorithm (speed pinned at 1.0x) runs once over the whole join, so the four parts read as a single video, and a random sound is laid over the top.

The sound runs from the start of the track until the footage ends, so the output always matches the full length of the four clips together — the track is trimmed there, and the video is never cut short. "Pick a Video…" pins the part-1 clip; parts 2-4, the order and the audio all stay random. No watermark.

Takes every screenshot in the frames folder, strips the background away and keeps only the text and icons on transparency. Run it once — the batch is stored on the server and reused by Fourcut V2 for every generate. Re-run it only when you add new screenshots or want to re-tune the bands.

The top and bottom values are how much of each screenshot holds UI. The middle is always dropped entirely. Defaults were tuned on the Tinder-style frames: 13% top keeps the status bar and tab row without catching the photo below it.

Or skip the folder entirely — pick screenshots straight off this computer and they get stripped and added to the batch. This one adds to what is already prepared instead of replacing it.

Composites a random input video into a random template screenshot: the template's top and bottom bands become the frame (fake app UI), the video fills the middle edge-to-edge with nothing drawn over it. Band heights are adjustable below — the defaults give the video the middle 50%. Then the repurposer runs with zoom and speed disabled so the UI edges and timing stay intact. "Pick a Video…" selects the input video; the template stays random.

Photo pool — stills only. Hit Browse Photos to see every pic in the folder and click one, or use the random button. The repurposer below is optional: when it's on it runs the pixel-level mods (brightness, contrast, saturation, hue, temperature, sharpness) plus a metadata strip and a fresh JPEG encode, so every download has a unique hash — but no zoom, no crop, no resize and no mirror, so the pic comes back framed exactly as uploaded. Turn it off to download the untouched original.

Same repurpose algorithm (10 randomised edits) over the Ruski library (dancing / lip-sync to rap). Picks one random video → metadata scramble → fresh-looking reel. No watermark.

Same repurpose algorithm (10 randomised edits) over the Reelsunbox library, then overlays the @reelsunbox tag at 70% opacity, bottom-center. Use this to eyeball the watermark placement.

Same repurpose algorithm as the main one, but pulls from a separate video library. Picks one random video → applies 10 randomised edits → fresh-looking reel.

Same repurpose pipeline as Test, but pulls from the CarTalk AI Drive folder. Picks one random video → applies 10 randomised edits → no watermark, no music — clean output ready to post.

Same repurpose pipeline, but pulls from the CarTalk REAL Drive folder — footage of the model actually talking on camera (no AI). Picks one random video → applies 10 randomised edits → clean output. Useful for posts where authentic human delivery matters more than scale.

Two options for the camera folder: Raw Photo gives the picture as-is (download + post manually anywhere). Repurpose Photo picks a random photo and applies the full randomised-edits chain (zoom, brightness, contrast, hue, saturation, mirror, color temperature, sharpness) — outputs a unique 1080×1920 JPG ready to post.

One Drive folder, both media types. Apxstraps Vid picks a random video → full repurpose chain (10 randomised edits) → mp4. Apxstraps Photo picks a random image → photo-only filter chain (brightness, contrast, hue, zoom, mirror, sharpness, colorbalance) → unique 1080×1920 JPG.

Clicking Combine V2 picks one random top and one random bottom video, and stacks them vertically — top half + bottom half, both playing at the same time.

Clicking Combine V3 will pick one random intro, one random main video, and optionally one end video from the V3 folders, join them together, and save the result.

Clicking Combine V3.1 uses the same V3 intro and end, but overlays the main video with a FaceTime-style PiP from the V3.1 overlay folder. Loading overlay count...

Clicking Combine V4 wraps a random intro and optional end around a middle segment where a random image sits in the top third and a random main video fills the bottom two-thirds.

Clicking Combine KIM wraps a random Kim intro (no captions) and optional end around a middle segment where a random image sits in the top third and a random Kim main video fills the bottom two-thirds.

Clicking Combine JOHNNY wraps a random Johnny intro (no captions) and optional end around a middle segment where a random image sits in the top third and a random Johnny main video fills the bottom two-thirds.

Clicking Combine BRAINROT composites a random overlay video (bottom half, centered) over a random full-screen background video. The background loops if shorter; output duration matches the overlay. Audio is taken from the overlay.

Clicking Repurpose OMEGLE picks a random video from the Omegle Reactions folder and applies 10 randomised visual transforms (zoom, brightness, contrast, saturation, hue, speed, noise, color temp, sharpness). Output goes to trial reels.

Clicking Repurpose Omegle+Intro picks a random intro from the Intro Pool, prepends it to a random Omegle reaction clip, then runs the same 10 randomised visual transforms + watermark as the plain Omegle pipeline. Output goes to trial reels (filename prefix omegle_intro_).

Clicking Repurpose BIKINIVIDS picks a random video from the BikiniVids folder and applies the standard randomised filter chain (zoom, brightness, contrast, saturation, hue, speed, noise, color temp, sharpness) to scramble the metadata. No watermark, no audio overlay.

Clicking Repurpose STORY VIDEO picks a random video from the Story Video folder and applies the standard randomised filter chain (zoom, brightness, contrast, saturation, hue, speed, noise, color temp, sharpness) to scramble the metadata. No watermark. Posted to IG Story by the persona loop when the dice rolls a video story.

Clicking Combine V5 picks one random main video and overlays a small FaceTime-style video in the top-right corner, then wraps it with intro and end.