# Image Studio

Image Studio is the design canvas inside the CoreReflex Studio — generate images with AI directly on the canvas at the size you need, retouch them with a mask brush, build layouts from text, shapes, and photos, and export print-ready files or drop the finished design straight onto your video timeline.

**On this page**

- [What is Image Studio?](#what-is-image-studio)
- [Why you'd use it](#why-youd-use-it)
- [Before you start](#before-you-start)
- [The canvas at a glance](#the-canvas-at-a-glance)
- [Create and set up a design](#create-and-set-up-a-design)
- [Generate an image with AI](#generate-an-image-with-ai)
- [AI editing](#ai-editing)
  - [Generative fill and object erase (the mask brush)](#generative-fill-and-object-erase-the-mask-brush)
  - [Expand an image to a new shape (outpaint)](#expand-an-image-to-a-new-shape-outpaint)
  - [Adjust the look of one image](#adjust-the-look-of-one-image)
  - [Ask the AI (conversational edits)](#ask-the-ai-conversational-edits)
- [Upscale an image 2×](#upscale-an-image-2)
- [Remove the background](#remove-the-background)
- [Filters for the whole design](#filters-for-the-whole-design)
- [Working with elements on the canvas](#working-with-elements-on-the-canvas)
- [Keyboard shortcuts](#keyboard-shortcuts)
- [Import files](#import-files)
- [Export your design](#export-your-design)
- [Place a design on the video timeline](#place-a-design-on-the-video-timeline)
- [Where your images live](#where-your-images-live)
- [Tips](#tips)
- [Troubleshooting](#troubleshooting)
- [Related pages](#related-pages)

---

## What is Image Studio?

Image Studio is a mode of the studio editor at [editor.corereflex.com](https://editor.corereflex.com) — the same place you edit video. Switch the mode toggle in the top bar from **Video** to **Image** and the timeline gives way to a design canvas: a fixed-size artboard where you layer images, text, and shapes, style them, and finish them with AI.

It shares everything with the rest of your project. The images in your project's media library appear in Image Studio ready to drop onto a design; every image you generate or edit here lands back in that same library; and a finished design can be placed onto the video timeline as an ordinary clip. One project, several rooms.

A few things make it different from a typical design tool:

- **You generate images *on* the canvas, to size.** Tell the AI what you want and where it should go — the whole canvas, a selected box, or exact pixel dimensions — and the result is placed for you.
- **AI edits are precise, not guesswork.** Fill and erase always work from a mask *you* paint with a brush, so the AI only ever touches the pixels you marked.
- **What you see is what you export.** The live canvas and every export run through the same drawing engine, so the downloaded file is pixel-identical to the preview — adjustments, filters, and all.
- **Nothing is ever destroyed.** Every AI edit, upscale, and background removal produces a *new* image; your original stays untouched in the media library, and every action can be undone with `Cmd/Ctrl+Z`.

## Why you'd use it

- Make social graphics, thumbnails, banners, and title cards in the same project as the video they promote.
- Generate exactly the image you need — right size, right spot — instead of generating, downloading, cropping, and re-uploading.
- Clean up photos: erase a photobomber, fill in a gap, extend a landscape shot to portrait, warm up the tones, cut out the subject.
- Produce a still overlay for your edit and drop it straight onto the timeline without leaving the app.

## Before you start

1. **Sign in.** Open [editor.corereflex.com](https://editor.corereflex.com). If AI actions tell you to sign in at corereflex.com first, sign in there — the studio and the dashboard share one account.
2. **Open a project.** Image Studio designs are saved with the project you have open, alongside its video edit and media.
3. **Know about credits.** The heavier AI actions — generating an image, fill/erase/expand edits, background removal — are metered in credits when your instance has metering switched on (image generation is 5 credits per run, a mask edit 6, background removal 4, and an "Ask the AI" request 3). Canvas work itself — arranging, text, shapes, adjustments, filters, exporting — costs nothing. See [Plans & Billing](plans-and-billing.md) for how credits work.

To enter Image Studio: click **Image** in the mode toggle at the top of the studio. If you've never used it in this project, an "Untitled design" canvas (1080×1080) is created for you automatically.

## The canvas at a glance

- **Center — the canvas.** Your artboard, automatically scaled to fit the window. A transparent background shows as a gray checkerboard. Click an element to select it; click empty space to deselect.
- **Right — the tools panel.** Top to bottom: the **Design** picker and canvas settings, **AI Generate**, **Add** (text / shape / upload, plus a grid of every image already in your project), the properties of whatever you have **Selected**, **AI Edit** (when an image is selected), **Filter** looks for the whole design, **Export**, and the **Place on timeline** button.

Everything you do — moving elements, AI results, filter changes — commits to the undo history. `Cmd/Ctrl+Z` steps back; your work also persists with the project as you go.

## Create and set up a design

A project can hold any number of designs (one canvas each).

1. **Pick or create a design.** The dropdown at the top of the panel lists every design in the project with its dimensions. Click **+ New** to add another (it starts at 1080×1080).
2. **Set the size.** Click a size preset to resize the current canvas:

   | Preset | Pixels | Typical use |
   |--------|--------|-------------|
   | 1:1 | 1080 × 1080 | Square social post |
   | 4:5 | 1080 × 1350 | Portrait feed post |
   | 9:16 | 1080 × 1920 | Story / vertical short cover |
   | 16:9 | 1920 × 1080 | Widescreen / thumbnail |
   | Banner | 1500 × 500 | Header / cover banner |

3. **Set the background.** Use the color swatch to pick a solid background, or click **Set transparent background** for a see-through canvas (the checkerboard). Transparency survives PNG and WebP export; JPG and PDF flatten it to white.

## Generate an image with AI

The **AI Generate** section turns a text description into an image placed on your canvas, sized to fit where you asked for it.

1. **Describe the image** in the prompt box — subject, style, lighting. The more specific, the better.
2. **(Optional) Click "+ Avoid…"** and list what you *don't* want — e.g. "text, watermarks, blur."
3. **Choose where it should go** with the size buttons:
   - **Canvas** — fill the whole artboard.
   - **Selection** — fill the box of the image or shape you currently have selected. (Disabled until you select one; text elements can't be a target.)
   - **Custom** — type an exact width × height in pixels (16–8192).
4. **Check the size readout.** Below the buttons you'll see the target box in pixels and the aspect ratio the image will be generated at (one of 1:1, 9:16, 16:9, 3:4, or 4:3 — the closest match to your box).
5. **Click Generate.** A progress note appears in the studio's activity bar; after a few seconds the image lands on the canvas as a new element, selected and ready to move or edit.

Worth knowing:

- The generator produces fixed aspect ratios, so the result is fitted *inside* your target box, centered, without stretching or cropping — if your box was an unusual shape you may see a small margin.
- When you generate into a **Selection**, the new image is placed *on top of* the selected element; the original stays underneath until you delete it.
- Every generated image is also saved to your project's media library, so you can reuse it in other designs or on the video timeline.
- The generated file is named from your prompt (e.g. "neon fox at dusk" → `neon-fox-at-dusk.png`) so it's easy to find later.

## AI editing

Select an image element on the canvas and the **AI Edit** section appears. It holds four tools: the mask brush (fill / erase), expand, and the conversational "Ask the AI" box. All of them produce a *new* image that replaces the picture on the selected element at the same visual size — the original file remains in your media library, and undo restores the element instantly.

### Generative fill and object erase (the mask brush)

Fill adds or replaces something *inside* the image; erase removes an object and reconstructs what was behind it. Both start from a mask you paint, so the AI only changes exactly what you marked.

1. **Select the image** on the canvas.
2. **Click "Fill / erase area"** in the AI Edit section. The canvas enters brush mode: your cursor becomes a crosshair with a circular brush outline, and a dashed border marks the image you're masking.
3. **Brush over the area** you want changed. Painted pixels show as a colored tint. While brushing:
   - **Brush size** — drag the slider (8–160 px).
   - **Paint / Unpaint** — switch to *Unpaint* to rub out parts of the mask you painted by mistake, then back to *Paint* to continue.
   - **Clear** — wipe the whole mask and start over.
   - You can paint in several strokes; they add up.
4. **For Fill:** type what should appear in the painted area (e.g. "a red umbrella") in the prompt box, then click **Fill**.
   **For Erase:** just click **Erase object** — no description needed; the AI removes whatever is under the mask and fills the hole to match the surroundings.
5. Wait a few seconds. The edited image replaces the picture on the element, at the same position and size. A confirmation ("Area filled" / "Object erased") appears.

Details that make this pleasant to use:

- **The mask sticks to the image.** Strokes are recorded relative to the picture itself, so if you move or rotate the element mid-masking, your painted area travels with it — nothing smears or drifts.
- **The Fill button stays disabled** until you've both painted something *and* written a prompt; Erase only needs paint. The button tooltips tell you which is missing.
- **Escape backs out.** Press `Esc` (or click **Cancel**) to leave brush mode without changing anything. Clicking a different element or switching designs also exits brush mode.
- While the brush is active, all mouse input paints — you can't accidentally drag or resize the element, and a stray Delete keypress won't remove it.

### Expand an image to a new shape (outpaint)

Expand grows an image beyond its current edges to a new aspect ratio — the AI continues the scene into the new space. Great for turning a landscape photo into a story cover, or a square into a widescreen banner.

1. **Select the image** on the canvas.
2. Under **Expand image**, click a target shape: **1:1**, **16:9**, **9:16**, **4:5**, or **4:3**.
3. Wait a few seconds. The element grows to the new shape with freshly generated surroundings — and your original pixels stay *exactly* where they were on the canvas, at exactly their original size. The new material appears around them.

If the image is already at the shape you clicked, nothing is charged or changed — you'll see "The image is already that aspect."

### Adjust the look of one image

Every image element carries its own non-destructive adjustment stack, separate from the whole-design filter. With an image selected, the **Adjust** sliders appear in the Selected section:

- **Brightness** (50–150%)
- **Contrast** (50–170%)
- **Saturation** (0–200%)
- **Warmth** (−60 to +60 — positive is warmer)
- **Shadows** (−100 to +100 — lift or crush the dark end)
- **Highlights** (−100 to +100)

Drag a slider and the canvas updates live; release it and the change commits as one undo step. Click **Reset** to clear all adjustments on that image. Because the preview and every export use the same rendering path, what you see while dragging is *exactly* what exports — no baking step, no surprises.

### Ask the AI (conversational edits)

At the bottom of AI Edit is a free-text box: describe the change in plain words and the assistant picks the right tool for you.

1. Select an image (for tone/fill/erase/expand asks) and type an instruction — for example:
   - *"make it warmer"* / *"more contrast"* / *"black and white"*
   - *"remove the car"*
   - *"add a sunset sky"*
   - *"make it widescreen"*
2. Click **Apply**.

What happens depends on the ask:

- **Tone and color changes apply instantly** to the image's adjustment stack — the assistant starts from your current slider values and moves them. It's one undo step, and a short note tells you what it did.
- **"Remove X" or "add X" arms the mask brush** with the description pre-filled — the assistant never guesses *where* in the picture you mean, so you brush over the spot and hit Fill or Erase yourself. A note walks you through it.
- **"Make it wider/taller/square" runs an expand** to the right aspect immediately.
- **Anything out of scope** gets a brief explanation of what the assistant *can* do instead.

Keep instructions under 500 characters; one change per ask works best.

## Upscale an image 2×

Any image element can be upscaled to double its resolution — useful before exporting large, or after several AI edits.

1. **Select the image** on the canvas.
2. **Click "Upscale 2×"** in the Selected section.

The enhanced image (higher-quality resample with sharpening, up to a maximum of 4096 px on the long edge) is saved to your media library as a *new* file — named like `photo-upscaled-2160w.png` — and the element on the canvas is switched onto it. The element keeps its visual size; it simply carries twice the pixels underneath, so it stays crisp when exported large or zoomed. If the image is already at the size cap, you'll see "Already at maximum resolution" and nothing changes.

## Remove the background

Cut the subject out of a photo and get a transparent PNG in one click.

1. **Select the image** on the canvas.
2. **Click "Remove background"** in the Selected section.

The subject is segmented and returned as a transparent-background cutout; the element on the canvas switches to it in place. The cutout is a new file — your original photo stays in the media library.

Two conditions to know about:

- **The image must be uploaded first.** Background removal works on the cloud copy of the file. If you just dropped a file in and it's still uploading, you'll see "Upload the image first" — give it a moment and try again.
- **Availability depends on your instance.** Background removal runs on your deployment's media-processing service. If that service isn't provisioned yet, the button reports that background removal is being provisioned and nothing is charged. Ask whoever operates your CoreReflex instance if you see this persistently.

> **Looking to turn a logo into a real vector (SVG)?** That's the **vectorize** tool in Brand Studio (the **Brand** mode in the same top-bar toggle), which traces a raster logo into clean SVG. Like background removal, it depends on your instance's media-processing service being provisioned.

## Filters for the whole design

The **Filter** section applies one photographic "look" across the entire design — every element, background included. Pick from twelve presets:

**Original · Vibrant · Muted · Warm · Cool · Dramatic · Fade · Cinematic · B&W Classic · B&W Punch · Vintage Film · Golden Hour**

…then fine-tune with the Brightness, Contrast, Saturation, and Warmth sliders beneath (adjusting a slider switches you off the preset onto your custom mix). The design look composes *under* any per-image adjustments, and — as everywhere in Image Studio — the export matches the preview exactly.

## Working with elements on the canvas

Designs are built from three kinds of elements: **images**, **text**, and **shapes** (rectangles). They stack in layers — later elements draw on top.

**Adding elements** (the **Add** section):

- **+ Text** drops an editable text block. With it selected you can type the copy, set the font size (8–400 px), pick the color, toggle **B**old and *I*talic, and align left / center / right. Text wraps automatically to the block's width.
- **+ Shape** drops a colored rectangle. Set its fill color and corner radius (0–200 px — max it out for a pill or circle).
- **Upload image / SVG** imports a file from your computer (see [Import files](#import-files)).
- The **image grid** below shows every image already in your project — generated stills, uploads, edit results. Click one to drop it onto the design, auto-fitted to the canvas.

**Selecting:**

- Click an element to select it. **Shift+click** adds or removes elements from the selection.
- Drag on empty canvas to **marquee-select** everything the rectangle touches.
- Click empty space (or press `Esc`) to deselect. `Cmd/Ctrl+A` selects everything.

**Moving, resizing, rotating** (single selection):

- **Drag** the element to move it. As you drag, pink **snap guides** appear when its edges or center line up with other elements or the canvas — cuts land clean without eyeballing.
- **Corner handles resize proportionally** — photos never skew. **Hold Shift while dragging a corner** to resize freely. **Edge handles** stretch one axis only. Resizing a text block from a corner scales its font size with it.
- **The round handle above the element rotates it**, pivoting around the center.

With **several elements selected**, dragging moves the whole group together; the panel offers group **Front / Back / Copy / Delete**.

**Everything else** (Selected section):

- **Opacity** slider (0–100%) on any element.
- **⬆ Front / ⬇ Back** move the selection to the top or bottom of the stack.
- **⧉ Copy** duplicates in place (offset slightly so you can see both).
- **Delete element** removes it (or press `Delete`).

## Keyboard shortcuts

Shortcuts apply when the canvas has focus — click the canvas first. (This also keeps them from colliding with the video editor's own shortcuts.)

| Shortcut | Action |
|----------|--------|
| `Cmd/Ctrl + C` | Copy the selected element(s) |
| `Cmd/Ctrl + V` | Paste (repeat pastes cascade so copies don't hide behind each other; works across designs) |
| `Cmd/Ctrl + D` | Duplicate the selection |
| `Cmd/Ctrl + A` | Select all elements |
| `Cmd/Ctrl + ]` | Bring the selection to the front |
| `Cmd/Ctrl + [` | Send the selection to the back |
| Arrow keys | Nudge the selection 1 px |
| `Shift + arrows` | Nudge 10 px |
| `Delete` / `Backspace` | Delete the selection |
| `Esc` | Deselect — or exit the mask brush |
| `Shift + click` | Add/remove an element from the selection |
| `Shift + drag corner` | Resize freely (unlocks the proportional constraint) |
| `Cmd/Ctrl + Z` | Undo (studio-wide) |

## Import files

Click **Upload image / SVG** in the Add section (PNG, JPG, WebP, GIF, SVG — any image your browser reads).

- **Bitmap images** are added to the design and to your project's media library, then uploaded to your CoreReflex storage in the background.
- **SVG files** are converted to a high-resolution PNG on import (up to 2048 px on the long side) so they behave like any other image on the canvas. If an SVG can't be read you'll see "Could not import SVG" and nothing is added.

Anything you upload here is also available in Video mode and every other design in the project.

## Export your design

The **Export** section downloads the current design as a file, pixel-identical to the canvas:

| Format | What you get |
|--------|--------------|
| **PNG** | Lossless raster; **keeps a transparent background** |
| **JPG** | Compact raster; transparency flattens to white |
| **WebP** | Modern compact raster; keeps transparency |
| **SVG** | A real vector file — text, shapes, and embedded images as scalable objects |
| **PDF** | A one-page document, print-ready (white backdrop) |
| **PPTX** | A one-slide PowerPoint with the design full-bleed — opens in PowerPoint, Keynote, or Google Slides |

Pick the format from the dropdown, click **Export**, and the file downloads to your device, named after the design. Exports render at the design's full pixel dimensions — the on-screen zoom never affects output quality.

## Place a design on the video timeline

The **Place on timeline** button at the bottom of the panel bridges Image Studio and the video edit:

1. Click **Place on timeline**.
2. The design is flattened to a full-resolution PNG — exactly as previewed — added to your media library, and dropped onto the video timeline as an image clip (2 seconds by default, front and center; the studio switches you to Video mode to see it).
3. Trim, move, and animate the clip like any other image.

The clip remembers where it came from: select it in Video mode and its inspector shows an **"Edit in Image Studio"** button that jumps you straight back to the editable design. The placed clip is a flattened snapshot, so after making changes, click **Place on timeline** again to drop an updated copy.

## Where your images live

Everything Image Studio produces flows into your project's shared media library, stored on your CoreReflex instance's own storage:

- **Generated images** — saved the moment they're created.
- **AI edit results** (fill, erase, expand) — saved as new files alongside the originals.
- **Upscales** — new files, clearly named with their new width.
- **Background-removal cutouts** — new transparent PNGs.
- **Uploads** — cached locally for instant display, uploaded to storage in the background.
- **Placed designs** — the flattened PNG joins the library too.

Originals are never overwritten; every operation creates a sibling file, so you can always step back. The one exception to "everything lands in the library" is the **Export** button, which downloads a file to your device.

## Tips

- **Generate into a selection** to produce an image at the exact spot and shape you need — draw a rectangle shape as a placeholder, select it, choose **Selection**, and generate.
- **Use "Avoid" liberally.** "text, watermark, extra fingers, blur" cleans up a lot of generations.
- **Erase before you fill.** Removing a distracting object first, then filling with what you actually want, usually beats one big fill.
- **Unpaint is your friend.** Overshot the mask? Switch the brush to Unpaint and tidy the edge instead of clearing and starting over.
- **Chain edits fearlessly.** Every result is a new file and one undo step — experiment, and `Cmd/Ctrl+Z` if it misses.
- **Upscale before exporting big.** If a design is headed for print or a large display, upscale its key images first so the export carries the extra detail.
- **Transparent PNGs layer beautifully on video.** Remove an image's background, place the design on the timeline over footage, and the cutout floats on the video.
- **Let "Ask the AI" drive.** If you're not sure which tool does what you want, describe the outcome — it picks the tool and sets it up.

## Troubleshooting

**"Sign in at corereflex.com first."**
Your studio session isn't signed in. Sign in on the dashboard, then retry — the two share one account.

**Generation or an edit fails with a credits message.**
Your workspace's credit balance is too low for that run (nothing was charged). Top up or upgrade under Settings → Plan — see [Plans & Billing](plans-and-billing.md).

**"Too many … requests — try again shortly."**
AI actions are rate-limited per user to keep the service healthy. Wait a minute and retry.

**"Upload the image first" when removing a background.**
The photo hasn't finished uploading to cloud storage yet (background removal works on the cloud copy). Wait for the upload to complete and click again.

**"Background removal is being provisioned and is not available yet."**
Your instance's media-processing service isn't deployed. No credits were spent. Contact whoever operates your CoreReflex deployment.

**The Fill button is grayed out.**
Fill needs two things: a painted mask *and* a description of what to generate. Erase only needs the mask. Hover the button — the tooltip says which is missing.

**"The image is already that aspect" when expanding.**
The image is already the shape you clicked; pick a different target aspect.

**"Already at maximum resolution" when upscaling.**
The image is at (or near) the 4096 px cap — it's already as large as Image Studio goes.

**I asked the AI to remove something and it told me to brush instead.**
That's deliberate: the assistant never guesses *which* pixels you mean. It arms the brush with your description pre-filled — paint the object and click Erase.

**My exported JPG/PDF has a white background but my canvas is transparent.**
JPG and PDF can't hold transparency, so it flattens to white. Export PNG or WebP to keep it.

**A generated image doesn't fill my custom box edge-to-edge.**
Images generate at the nearest standard aspect ratio and are fitted inside your box without stretching. Use a box close to 1:1, 4:3, 3:4, 16:9, or 9:16 for a perfect fill — or expand the result afterward.

**My brush strokes vanished.**
Clicking a different element or switching designs exits mask mode (the mask belongs to one image). Reselect the image and start the brush again.

## Related pages

- [Getting Started](getting-started.md) — sign in, accounts, and your first project.
- [Editing & Mastering](editing-and-mastering.md) — the video timeline your placed designs land on.
- [Writing & Content](writing-and-content.md) — copy and script tools that pair with your visuals.
- [Plans & Billing](plans-and-billing.md) — how credits and plan tiers work.