Skip to main content

Getting Started with ZenCreator

Create a Persona, keep it consistent across photos & video, and start posting today.

Updated over a month ago

The Big Idea

ZenCreator is a studio in a browser. You’ll generate photos and short videos, keep a single Persona consistent across everything, and ship content to your socials. Our native publishing and scheduling arrive in Version 2; until then, you’ll export and post manually (workflow below).

TL;DR (10 minutes to first wins)

  1. Create your Persona face with AI Face Generation → pick a look you love.

  2. Lock identity using Face Swap (run last) or the combined Upscale + FaceSwap flow.

  3. Produce sets:

  4. Polish with Upscaler (choose a Safe Face mode when identity matters).

  5. Export and publish manually (guidelines below). Done.

Step 1 — Craft your Persona (identity you’ll carry everywhere)

  • Start in Face Generation. Use Surprise Me if you want inspiration, or set traits manually (gender, age, ethnos, hair, beard/makeup, etc.).

  • Each run returns 1-4 portrait. Shortlist 1–2 “foundation” faces.

  • Tip: Collect reference images (outfits, moods, locations). You’ll use these in Generation by Ref and PhotoShoot.

Consistency rule: your Persona is a brand. Keep hair length/color, makeup level, and overall vibe stable across shoots unless you deliberately plan a change (new “season”).

Step 2 — Lock identity (so every image looks like the same person)

  • Generate your scenes first, then Face Swap last onto your Persona for pixel-level identity.

  • If you already know the final face, save clicks with the Upscale + FaceSwap combined flow: it upscales and swaps in one pass—our near-best default for most pipelines.

  • When you only need bigger images and face must not change, choose Upscaler → Basic Safe Face (fast) or Premium Safe Face (highest realism).

Step 3 — Pick the right generator for the job

PhotoShoot — one run, many categories (office, café, sport…). Ideal for building a week of posts fast.
Generation by Reference — upload a photo you like; get similar shots (pose, framing, lighting) with variations. Add LoRAs on SDXL if you want physique/styling control.
Generation by Prompt — text-to-image (models: WAN realism or SDXL + LoRAs). Use when you want ideas from scratch.
Carousel — turn a single photo into a swipeable set of angles and crops that feel cohesive.
Video Generation — animate a still into a 5–10s clip.

  • Kling 2.1 → most realistic; beauty/fashion/product hero shots.

  • Kling 1.6 → fast, clean, reliable batches (5s/10s).

  • WAN → edgier looks, looser motion.
    Prompt for subtle, believable motion: “gentle breathing, soft eye blink, slow 3% dolly-in, hair barely moving, neutral film grade.”
    Keep negatives tight: flicker, ghosting, plastic skin, heavy sharpening, hair across eyes.

Step 4 — Polish with Upscaler (which mode?)

  • Basic: fast detail, may change face (drafts/bulk).

  • Basic Safe Face: fast and keeps the face unchanged (great default).

  • Premium Realism: highest fidelity, may change face.

  • Premium Safe Face: highest fidelity and keeps the face.

  • Sharp Keep: strict sharpness with minimal change (e-commerce).

  • Textured Boost: adds realistic pores/fabric texture when images feel “plastic”.

Rule of thumb: Safe Face when identity matters; Premium when final beauty matters; Sharp Keep for products.

Step 5 — Build a “shoot” like a real team

  1. Choose an aspect ratio for the whole set (IG feed 4:5, Stories/Reels 9:16).

  2. Batch with PhotoShoot or Gen by Ref (8–12 variations per scene).

  3. Curate: pick winners, then Upscale (Safe Face) → Face Swap last.

  4. Video: animate the hero still (Kling 2.1, 5s). Create 3–5 clips with one motion idea each.

  5. Carousel: extract 5–8 frames that feel like one story.

  6. Export: download selected or all.

Step 6 — Publish now

Until v2, use this simple manual loop:

  • Instagram Feed: 1080×1350 (4:5). One strong image or a carousel; short caption + 1–3 specific hashtags.

  • Instagram Reels / TikTok / Shorts: 1080×1920 (9:16). Keep to 6–12s for higher completion. Add music in-app.

  • Twitter/X & Threads: 1200×1600 (portrait) or 1600×900 (landscape). Keep copy punchy.

  • YouTube Community: 3:2 image or 9:16 short clip; frame the face big.

Workflow: export → airdrop/drive → upload natively. Save your caption templates in a notes doc for reuse.


Pro move: make a weekly Content Checklist (“2 Reels, 4 photos, 1 carousel”), and batch-create everything Sunday night.

Step 7 — Keep your Persona consistent over weeks

  • Reference board: pin hair/makeup/outfit looks you approve; reuse them.

  • Seeds & settings: when you love a look, reuse seeds and LoRA strengths.

  • Color grade: choose a palette (warm/neutral/cool) and keep it across sets.

  • Angles: 2–3 recurring angles become visual signature (e.g., 3/4 headshot, mid-shot, OTS mirror).

  • Naming: prefix files with persona (maya_…), keep aspect (4x5/9x16) and date.

A practical first week plan

Day 1: Face Generation → pick Persona → create a 10-image PhotoShoot.
Day 2: Curate → Upscale (Safe Face)Face Swap → export 6 winners.
Day 3: Video Gen: 3 clips (5s each) from the hero image (Kling 2.1).
Day 4: Carousel from the hero set (5–8 slides).
Day 5: Gen by Ref on a favorite pose for 8 more picks.
Day 6–7: Post manually to IG/TikTok/Threads; record metrics (saves/comments). Repeat what works.

Common pitfalls (and easy fixes)

  • “Why does the face change after upscaling?” Use Safe Face modes or the combined Upscale + FaceSwap flow.

  • “Videos look plasticky.” Add “realistic skin texture, natural pores” to prompt; use Kling 2.1; avoid extreme motion.

  • “Hands look weird.” Crop tighter, freeze hands (no gestures), add bad hands, finger warp to negatives.

  • “Inconsistent vibe.” Fix aspect ratio, reuse seeds, keep one color grade for the whole batch.

  • “Too much work to swap faces after.” Use the combined Upscale + FaceSwap flow to finish in one pass.

FAQs

Do I own the images/videos? Yes—royalty-free for commercial use.
How many images do tools output? Vary by your settings.
Can I upload a real person’s face? Yes—but always respect consent and local laws.
Where do I see all my outputs? Tools; from any result grid you can download or send to another tool.

Did this answer your question?