object.zero__
Abstract 3D posts for Instagram
Post targets
Automation schedule
Posts are spaced through your day with random jitter. One queue item runs per minute when due. Plan refills daily at 00:05 (your timezone). Reels use the uploaded video staged below. Requires the app running 24/7 or a host in the cloud.
Queue (today & tomorrow)
Times shown in your schedule timezone.
| Date | Local time | Type | Status |
|---|---|---|---|
| Loading… | |||
Reel video queue
Upload one or many MP4/MOV files (max 100 MB each). Videos are queued in upload order — the scheduler posts the next pending video at each Reel slot (Reels per day). Example: 20 videos + 1 reel/day ≈ 20 days of posts. Files live on the server under uploads/videos/. Caption + hashtags come from the Caption section.
| # | File | Status | Action |
|---|---|---|---|
| Loading… | |||
Caption & Hashtags
Feed posts get a rotating caption (80% one word, 20% a question) plus hashtags. Stories have no caption; Reels get hashtags only.
Rotating feed caption
Last word used: — (same word won’t repeat back-to-back)
—
Word pool (80%)
Question pool (20%)
Hashtags: always tags on every post; pool tags chosen at random within min/max each time.
Dynamic abstract generation
Production pipeline: anchor edit with 4 form families, gloss/color reference, light BG rotation, anti-repeat (last 7). Intensity controls below modulate gloss, color, form variation, surreal, and edge in the anchor pipeline.
Dynamic intensities
Steuern Gloss, Farbe, Form-Variation, Surreal und Kante in der Anchor-Edit-Pipeline. Save klicken vor Generate Preview.
Modulation 1: ±0,1…±0,9 · Modulation 10: ±0,1…±9,0 (0,1-Schritte). Pro Parameter eigene Checkbox und Modulations-Regler.
Low = compact/folded. High = elongated, twisted.
Low = subtle. High = bold impossible geometry.
Low = weiche Silhouette. High = lesbare Kontur.
0.01 = opak glossy. 10 = milchig-transluzent.
Low = sanfte Töne. High = kräftigere, gesättigtere Farbkombinationen (2–4 Farben).
Prompt library (legacy)
Used only when dynamic abstract is off for that channel, or for reference. Check In pool for random selection with “Random from pool”. Changing the radios saves immediately.
Feed — image prompt source
Story — image prompt source
Saved prompts
| Type | Label | Pool | Prompt | Actions |
|---|---|---|---|---|
| Loading… | ||||
Add prompt
Image Prompt
Manual feed prompts (used when “Feed — manual” is selected, or as fallback if the pool is empty). If “Random from pool” is selected, Generate preview uses the pool the same way as scheduled posts—not these text fields.
Model A/B test (ChatGPT reference)
Compare reference master vs dynamic pipeline (variants forced on for the right side, even in Phase 1). Both use gpt-image-1 when dall-e-3 is unavailable. ~2× API cost.
After Generate preview, that file is staged for Post Feed now only. Scheduled automation ignores it and always builds a new image from pool/manual prompts (so you are not stuck on one picture).
Story image
Manual story prompts (used when “Story — manual” is selected, or as fallback). If “Random from pool” is selected, Generate Story Preview uses the pool like automation. Staged story previews apply to Post Story now only; the queue always generates fresh from pool/manual.
Manual post
Last 20 posts
| Time | Type | Status | ID / Message |
|---|---|---|---|
| Loading… | |||