> For the complete documentation index, see [llms.txt](https://docs.qolaba.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.qolaba.ai/qolaba-2.0/video-series/scene-by-scene-flow.md).

# "Scene by Scene" Flow

This flow gives you full manual control. Instead of describing a single idea and letting the AI plan the scenes, you start with a blank canvas and write each scene prompt yourself. This is ideal when you have a specific shot list, storyboard, or creative brief that you want to follow precisely.

#### Selecting Scene by Scene mode

On the Video Series home screen, click the **Scene by scene** mode card on the right. The subtitle reads: “Blank scenes, you fill them in.” Set your orientation and duration, then click the orange **Start blank** button.

![Video Series home with 'Scene by scene' card selected, showing 'Blank scenes, you fill them in' subtitle, configuration pills for Omni video model, Portrait, Landscape, 10s, and orange 'Start blank' button](https://content.gitbook.com/content/EMTAJyXREpSSpII9h335/blobs/BmQhaG8JAdnBNnIheqAE/Unknown%20image)

The **Scene by scene** card is selected (outlined in orange). The prompt area is replaced by configuration pills and a **Start blank** button. Templates are still visible below for quick access.

#### The blank canvas

You land on the Scenes stage (Step 1) with a single empty scene card. The interface has two main sections:

* **Carried through every scene**: A reference image area at the top where you upload images of your character, product, or subject. These images maintain visual consistency across all generated clips.
* **Your scenes**: Below, numbered scene cards with text areas where you write each scene's prompt. A **+ Add a scene** button lets you add more.

![Blank scene editor showing 'Scenes' heading, a banner reading 'Enjoy the free beta with unlimited clips!', the 'Carried through every scene' reference image area with an '+ Add image' placeholder, one empty Scene 1 card with 'Describe this scene...' placeholder, and footer showing '0 clips - Free' with '+ Add a scene' and 'Generate Free' buttons](https://content.gitbook.com/content/EMTAJyXREpSSpII9h335/blobs/q5CkQksoMfSGOMqvTgES/Unknown%20image)

A fresh Scene by scene canvas. The reference image area (“Carried through every scene”) is empty, ready for you to upload a character or product photo. Scene 1 has a blank text area. The footer confirms 0 clips and free generation.

#### Adding reference images

Click the **+ Add image** placeholder in the “Carried through every scene” section. A modal opens where you can drag-and-drop an image (up to 20 MB) or browse your files. You can also choose from your generation history, filtered by All, Generated, or Uploaded.

![Add reference image modal showing drag-and-drop area with 'Maximum size: 20MB', a 'Choose from history' section with All/Generated/Uploaded tabs, and a grid of previously generated and uploaded images](https://content.gitbook.com/content/EMTAJyXREpSSpII9h335/blobs/ojZOvol4GhSjQH1IpbXc/Unknown%20image)

The reference image picker. Upload a new file or choose from your existing image history. The tabs let you filter between AI-generated images and manual uploads.

#### Writing scene prompts with reference images

After adding reference images, each one gets a name label (editable) and appears in the “Carried through every scene” section. You can add multiple reference images, for example, a character photo labelled “Sam” and a product photo labelled “Classic watch.” These references are automatically incorporated into every scene's generation to maintain visual consistency.

![Scene editor with two reference images uploaded: a portrait photo labelled 'Sam' and a watch photo labelled 'Classic watch', plus an '+ Add image' slot. Below, four scene cards with detailed prompts describing Sam discovering a vintage watch, nostalgic flashback, returning to present, and walking down a street](https://content.gitbook.com/content/EMTAJyXREpSSpII9h335/blobs/uHcV73LtKhODOKJ9Jeb7/Unknown%20image)

A fully populated scene editor. Two reference images (“Sam” and “Classic watch”) are set to carry through every scene. Four scenes describe a narrative arc about discovering a vintage watch. Each scene prompt includes cinematic direction: camera movement, lighting, depth of field, and film grain.

#### Editing reference image descriptions

Qolaba automatically generates a text description of each uploaded reference image using AI vision. You can edit this description to refine what the model focuses on, skin tone, clothing, distinguishing features, product details. Click the pencil icon next to the image label to open the description editor.

#### Generating and reviewing clips

Once your scenes are written and reference images are set, click the orange **Generate** button. Qolaba renders each scene into an individual video clip. The pipeline stepper advances to Step 2 (Clips), and you see all generated clips in a grid with status badges.

![Clips stage showing four generated video clips in a grid, each with a green Done badge, Regenerate Free button, and scene description preview. Scene 3 shows Takes 1 and 2.](https://content.gitbook.com/content/EMTAJyXREpSSpII9h335/blobs/teHUtqCGXQG4sTMsThgV/Unknown%20image)

All four clips have generated (4/4 done). Each clip card shows a video thumbnail with playback controls, a green **Done** badge, and a **Regenerate** button. Scene 3 already has two Takes (indicated by the numbered circles), showing it was regenerated for a better result. Click **Assemble video →** to proceed to the final step.

#### Regenerating in Scene by Scene

The regeneration workflow is identical to the “I have an idea” flow. Click **Regenerate** on any clip to open the Edit Scene modal, where you can refine the prompt and create a new take.

![Edit Scene 1 modal in Scene by Scene flow with clip preview, editable prompt about Sam discovering a watch, info banner, and Regenerate button](https://content.gitbook.com/content/EMTAJyXREpSSpII9h335/blobs/XKnuWMHviAWVWDncRpgT/Unknown%20image)

The Edit Scene modal for Scene 1 in the Scene by Scene flow. The prompt contains the full cinematic description you wrote manually. You can refine it, adjust camera angles, change the lighting mood, add detail, then click **Regenerate**. Your original clip is preserved as a separate take.

#### Setting transitions in Scene by Scene

On the Assemble stage, click any gap between clips in the timeline to open the Transition Preview. The same seven transition types are available across all flows.

![Transition Preview modal in Scene by Scene flow showing live preview, seven transition type buttons with Hard Cut selected, and Apply to all scenes button](https://content.gitbook.com/content/EMTAJyXREpSSpII9h335/blobs/r2zx1x4O42UENHq3OK7W/Unknown%20image)

The Transition Preview modal in the Scene by Scene flow. A live preview shows the actual footage from your clips with the selected transition applied. Choose a type, preview it, and click **Apply to all scenes** to set the same transition everywhere, or set each gap individually.