2Shot Scene _ SX CU — v1.0
🎬 Cinematic 2‐Shot Scene Generator – ComfyUI Workflow
Profil d'exécution
- Accès
- Flux de travail externe
- Statut
- Présenté
- Version
- cinematic2ShotScene_v10.json
- Source
- Civitai
- Mis à jour
- 14 oct. 2026
Description de la source
🎬 Cinematic 2‐Shot Scene Generator – ComfyUI Workflow
Created by Me – a filmmaker & AI workflow designer. This is my all‐in‐one tool for generating professional cinematic two‐shot scenes from just three input images: a background and two character portraits . It automatically removes backgrounds, composes the characters into the scene, and produces 12 different camera angles – from close‐ups to wide shots, low/high angles, and side views – all in a single run.
🧩 Overview
This advanced workflow takes a background image and two character photos, removes the characters’ backgrounds using BiRefNet, composites them into the scene, and then generates a full set of camera angles (Close‐up two‐shot, Medium two‐shot, Wide, Ultra‐wide, 30°/60° left/right, Low angle, High angle, and face close‐ups for each character). It also includes an optional auto‐prompt generator that uses Gemma 4 to analyse your images and build a detailed text description, or you can switch to a fully text‐based generation mode.
1️⃣ Required Models & Prerequisites
Make sure these files are placed in your ComfyUI models/ folders:
| Folder | Required Files |
|--------|----------------|
| diffusion_models/ | qwen_image_edit_2511_bf16.safetensors (or fp8 version) |
| loras/ | Qwen-Image-Edit-2511-Lightning-4steps-V1.0-bf16.safetensors qwen-image-edit-2511-multiple-angles-lora.safetensors aldniki_qwen_reality_transform_v01.safetensors (optional for realism) |
| text_encoders/ | qwen_2.5_vl_7b_fp8_scaled.safetensors gemma4_e4b_it_fp8_scaled.safetensors (for auto‐prompt) |
| vae/ | qwen_image_vae.safetensors |
| bgremoval/ | birefnet.safetensors (for automatic background removal) |
⚠️ VRAM Requirement: At least 12–16 GB recommended, as the workflow loads multiple models (Qwen, Gemma, BiRefNet) simultaneously.
3️⃣ How It Works (Automatic Composition & Multi‐Angle Generation)
-
Load your images – one background (location) and two character portraits.
-
Background removal – BiRefNet automatically removes the backgrounds from both character images.
-
Composition – the cleaned characters are placed onto the background scene using ImageCompositeMasked nodes.
-
Auto‐prompting (optional) – Gemma 4 analyses the images and generates a detailed prompt describing the scene and characters. You can also provide your own custom prompt.
-
Multi‐angle generation – the composed scene is fed into 12 separate KSampler pipelines (each with the Multiple‐Angles LoRA), producing:
-
Close‐up two‐shot
-
Medium two‐shot
-
Wide
-
Ultra‐wide
-
30° left / 30° right
-
60° left / 60° right
-
Low angle
-
High angle
-
Close‐up face left / Close‐up face right
- All outputs are saved as a batch (or individually) with clear prefixes.
6️⃣ Advanced Settings (Optional)
-
Resolution – change the output size in the EmptySD3LatentImage node (default: 1920×1088).
-
Image‐based vs Text‐based – a switch lets you choose between composing from the actual images off ) or generating the scene purely from the text description on ).
-
LoRA Strength – adjust the reality_transform and lightning LoRA weights (default = 1) to fine‐tune the style.
-
Auto‐Prompt Length – control the response length of the Gemma‐based prompt generator.
-
Model Selection – the workflow uses Qwen‐Image‐Edit 2511 with Lightning LoRA for speed; you can replace it with any other compatible model.
📦 Final Output
After running, you’ll get two sets of images saved in the Qwen _ 2Shot Scene/ folder:
-
SX_Base_* – the basic composed scene (base image).
-
2Shot_* – a batch containing all 12 camera angles in one file (or individual files if you save separately).
Each angle is generated with the specified camera parameters and saved with a descriptive name.
Why I built this: As a filmmaker, I often need to visualise different shot compositions for a scene without reshooting or manual editing. This workflow gives me a complete set of angles in minutes – perfect for pre‐visualisation, storyboarding, or even final shots. I hope it saves you time and opens up new creative possibilities! 🚀🎥
📬 Contact: If you’d like to donate, ask a question, or order a custom workflow, reach out to me on Civitai (SaeedKhalili) , visit my website SaeedKhalili.com , or email me at [email protected] . I’m always happy to help fellow creators!
What you need
- Model files
- 9.9 GiB known subtotal
Remaining hardware or download-size estimates are not available yet.
Exigences
Modèles
8aldniki_qwen_reality_transform_v01.safetensors
Non résoluLoRA · Unknown
- Trouvé
Checkpoint · 424 MB · SAFETENSORS · Unknown
gemma4_e4b_it_fp8_scaled.safetensors
Fichier non vérifiéText encoder · Hugging Face · Comfy-Org/gemma-4 · Gemma 4
Text encoder · 8.74 GB · Hugging Face · Comfy-Org/Qwen-Image_ComfyUI
qwen_image_2512_fp8_e4m3fn.safetensors
Non résoluCheckpoint · Hugging Face · Comfy-Org/Qwen-Image_ComfyUI
qwen_image_edit_2509_fp8_e4m3fn.safetensors
Fichier non vérifiéCheckpoint · Hugging Face · Comfy-Org/Qwen-Image-Edit_ComfyUI · Comfy-Org/Qwen-Image-Edit_ComfyUI · Hugging Face
- Fichier non vérifié
VAE · Hugging Face · Comfy-Org/Qwen-Image_ComfyUI
LoRA · 810 MB · Hugging Face · lightx2v/Qwen-Image-Lightning
Sources
1 source
Sources
1 sourceExtraits de sources
1 extrait
Extraits de sources
1 extraitSource context: 77 downloads · Type Workflows · Base model Qwen
