release: v1.2.0

## Summary

- Updates the reference image to the latest version
- Removes the barefoot/no socks instruction from the character prompt
- Clarifies that vampire fangs are only visible when the mouth is open

✨ This PR was created with help from Hikari~ 🌸

Reviewed-on: #20
Co-authored-by: Hikari <hikari@nhcarrigan.com>
Co-committed-by: Hikari <hikari@nhcarrigan.com>

package.json

@@ -1,7 +1,7 @@
 {
   "name": "tatsumi",
   "private": true,
-  "version": "1.0.0",
+  "version": "1.2.0",
   "type": "module",
   "scripts": {
     "dev": "vite",
@@ -23,7 +23,7 @@
     "@tauri-apps/cli": "2.5.0",
     "@types/react": "19.1.2",
     "@types/react-dom": "19.1.2",
-    "autoprefixer": "10.4.27",
+    "autoprefixer": "10.4.21",
     "eslint": "9.25.1",
     "postcss": "8.5.3",
     "tailwindcss": "3.4.17",

pnpm-lock.yaml

@@ -43,8 +43,8 @@ importers:
         specifier: 4.4.1
         version: 4.4.1(vite@6.3.2(jiti@1.21.7)(yaml@2.8.3))
       autoprefixer:
-        specifier: 10.4.27
-        version: 10.4.27(postcss@8.5.3)
+        specifier: 10.4.21
+        version: 10.4.21(postcss@8.5.3)
       eslint:
         specifier: 9.25.1
         version: 9.25.1(jiti@1.21.7)
@@ -935,8 +935,8 @@ packages:
     resolution: {integrity: sha512-hsU18Ae8CDTR6Kgu9DYf0EbCr/a5iGL0rytQDobUcdpYOKokk8LEjVphnXkDkgpi0wYVsqrXuP0bZxJaTqdgoA==}
     engines: {node: '>= 0.4'}
 
-  autoprefixer@10.4.27:
-    resolution: {integrity: sha512-NP9APE+tO+LuJGn7/9+cohklunJsXWiaWEfV3si4Gi/XHDwVNgkwr1J3RQYFIvPy76GmJ9/bW8vyoU1LcxwKHA==}
+  autoprefixer@10.4.21:
+    resolution: {integrity: sha512-O+A6LWV5LDHSJD3LjHYoNi4VLsj/Whi7k6zG12xTYaU4cQ8oxQGckXNX8cRHK5yOZ/ppVHe0ZBXGzSV9jXdVbQ==}
     engines: {node: ^10 || ^12 || >=14}
     hasBin: true
     peerDependencies:
@@ -1364,8 +1364,8 @@ packages:
     resolution: {integrity: sha512-dKx12eRCVIzqCxFGplyFKJMPvLEWgmNtUrpTiJIR5u97zEhRG8ySrtboPHZXx7daLxQVrl643cTzbab2tkQjxg==}
     engines: {node: '>= 0.4'}
 
-  fraction.js@5.3.4:
-    resolution: {integrity: sha512-1X1NTtiJphryn/uLQz3whtY6jK3fTqoE3ohKs0tT+Ujr1W59oopxmoEh7Lu5p6vBaPbgoM0bzveAW4Qi5RyWDQ==}
+  fraction.js@4.3.7:
+    resolution: {integrity: sha512-ZsDfxO51wGAXREY55a7la9LScWpwv9RxIrYABrlvOFBlH/ShPnrtsXeuUIfXKKOVicNxQ+o8JTbJvjS4M89yew==}
 
   fsevents@2.3.2:
     resolution: {integrity: sha512-xiqMQR4xAeHTuB9uWm+fFRcIOgKBMiOBP+eXiyT7jsgVCq1bkVygt00oASowB7EdtpOHaaPgKt812P9ab+DDKA==}
@@ -1755,6 +1755,10 @@ packages:
     resolution: {integrity: sha512-6eZs5Ls3WtCisHWp9S2GUy8dqkpGi4BVSz3GaqiE6ezub0512ESztXUwUB6C6IKbQkY2Pnb/mD4WYojCRwcwLA==}
     engines: {node: '>=0.10.0'}
 
+  normalize-range@0.1.2:
+    resolution: {integrity: sha512-bdok/XvKII3nUpklnV6P2hxtMNrCboOjAcyBuQnWEhO665FwrSNRxU+AqpsyvO6LgGYPspN+lu5CLtw4jPRKNA==}
+    engines: {node: '>=0.10.0'}
+
   object-assign@4.1.1:
     resolution: {integrity: sha512-rJgTQnkUnH1sFw8yT6VSU3zD3sWmu6sZhIseY8VX+GRu3P6F7Fu+JNDoXfklElbLJSnc3FUQHVe4cU5hj+BcUg==}
     engines: {node: '>=0.10.0'}
@@ -3279,11 +3283,12 @@ snapshots:
 
   async-function@1.0.0: {}
 
-  autoprefixer@10.4.27(postcss@8.5.3):
+  autoprefixer@10.4.21(postcss@8.5.3):
     dependencies:
       browserslist: 4.28.2
       caniuse-lite: 1.0.30001787
-      fraction.js: 5.3.4
+      fraction.js: 4.3.7
+      normalize-range: 0.1.2
       picocolors: 1.1.1
       postcss: 8.5.3
       postcss-value-parser: 4.2.0
@@ -3872,7 +3877,7 @@ snapshots:
     dependencies:
       is-callable: 1.2.7
 
-  fraction.js@5.3.4: {}
+  fraction.js@4.3.7: {}
 
   fsevents@2.3.2:
     optional: true
@@ -4252,6 +4257,8 @@ snapshots:
 
   normalize-path@3.0.0: {}
 
+  normalize-range@0.1.2: {}
+
   object-assign@4.1.1: {}
 
   object-hash@3.0.0: {}

src-tauri/Cargo.lock

@@ -3716,7 +3716,7 @@ checksum = "61c41af27dd6d1e27b1b16b489db798443478cef1f06a660c96db617ba5de3b1"
 
 [[package]]
 name = "tatsumi"
-version = "1.0.0"
+version = "1.2.0"
 dependencies = [
  "base64 0.22.1",
  "dirs 5.0.1",

src-tauri/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tatsumi"
-version = "1.0.0"
+version = "1.2.0"
 description = "Tatsumi - AI art generation using Google Gemini"
 authors = ["Naomi Carrigan"]
 edition = "2021"

src-tauri/src/gemini.rs

@@ -11,11 +11,11 @@ Character design (always required):
 - Wavy ashen brown hair (colour and texture fixed; hairstyle can vary)
 - Very pale skin tone
 - Vibrant sky-blue eyes — important, commonly missed
-- Vampire fangs
+- Vampire fangs (only visible when mouth is open)
 - Glasses (pink-framed preferred, other styles acceptable)
 - Painted fingernails and toenails (any colour, never unpolished)
 - Slender build
-- Full body visible in frame; always barefoot, never wears socks
+- Full body visible in frame
 
 Composition (always required):
 - Single character only
@@ -46,12 +46,19 @@ fn build_safety_settings() -> Value {
     ])
 }
 
-fn build_generation_config(mode: &str) -> Value {
+fn build_generation_config(
+    mode: &str,
+    aspect_ratio: Option<&str>,
+    image_size: &str,
+) -> Value {
     let image_config = match mode {
-        "avatar" => json!({ "aspectRatio": "1:1", "imageSize": "4K" }),
-        "art"    => json!({ "aspectRatio": "16:9", "imageSize": "4K" }),
+        "avatar" => json!({ "aspectRatio": "1:1", "imageSize": image_size }),
+        "art" => {
+            let ratio = aspect_ratio.unwrap_or("16:9");
+            json!({ "aspectRatio": ratio, "imageSize": image_size })
+        }
         // replace mode: omit aspectRatio so the model infers it from the source image
-        _        => json!({ "imageSize": "4K" }),
+        _ => json!({ "imageSize": image_size }),
     };
     json!({
         "imageConfig": image_config,
@@ -83,37 +90,56 @@ fn build_user_gemini_parts(
     user_image_base64: &Option<String>,
     user_image_mime: &Option<String>,
 ) -> Vec<Value> {
-    if mode == "replace" && user_image_base64.is_some() {
+    if let Some(image_data) = user_image_base64.as_deref() {
         let mime = user_image_mime.as_deref().unwrap_or("image/png");
-        let data = user_image_base64.as_deref().unwrap_or("");
         let base_text = user_text.as_deref().unwrap_or("");
-        let final_text = if base_text.is_empty() {
-            REPLACE_MODE_APPEND.to_string()
+        let final_text = if mode == "replace" {
+            if base_text.is_empty() {
+                REPLACE_MODE_APPEND.to_string()
+            } else {
+                format!("{}\n{}", base_text, REPLACE_MODE_APPEND)
+            }
         } else {
-            format!("{}\n{}", base_text, REPLACE_MODE_APPEND)
+            base_text.to_string()
         };
 
-        vec![
-            json!({"inlineData": {"mimeType": mime, "data": data}}),
-            json!({"text": final_text}),
-        ]
+        let mut parts = vec![json!({"inlineData": {"mimeType": mime, "data": image_data}})];
+        if !final_text.is_empty() {
+            parts.push(json!({"text": final_text}));
+        }
+        parts
     } else {
-        // Art/avatar mode, or replace mode follow-up correction (text only)
+        // No image: text-only message
         let text = user_text.as_deref().unwrap_or("");
         vec![json!({"text": text})]
     }
 }
 
+pub struct GeminiCallParams {
+    pub mode: String,
+    pub aspect_ratio: Option<String>,
+    pub image_size: String,
+    pub user_text: Option<String>,
+    pub user_image_base64: Option<String>,
+    pub user_image_mime: Option<String>,
+}
+
 pub async fn call_gemini(
     api_key: String,
-    mode: String,
     history: Vec<ThreadMessage>,
-    user_text: Option<String>,
-    user_image_base64: Option<String>,
-    user_image_mime: Option<String>,
+    params: GeminiCallParams,
 ) -> Result<(Vec<MessagePart>, f64), String> {
     let client = reqwest::Client::new();
 
+    let GeminiCallParams {
+        mode,
+        aspect_ratio,
+        image_size,
+        user_text,
+        user_image_base64,
+        user_image_mime,
+    } = params;
+
     let is_first_message = history.is_empty();
 
     let mut contents: Vec<Value> = history
@@ -156,7 +182,11 @@ pub async fn call_gemini(
 
     contents.push(json!({"role": "user", "parts": user_parts}));
 
-    let generation_config = build_generation_config(mode.as_str());
+    let generation_config = build_generation_config(
+        mode.as_str(),
+        aspect_ratio.as_deref(),
+        image_size.as_str(),
+    );
     let safety_settings = build_safety_settings();
 
     let request_body = json!({
@@ -248,8 +278,13 @@ pub async fn call_gemini(
     let candidates_tokens = usage["candidatesTokenCount"].as_u64().unwrap_or(0);
     let image_part_count = result_parts.iter().filter(|p| p.part_type == "image").count() as u64;
 
-    // Image output tokens (4K = 2000 tokens each) billed at $120/1M tokens
-    let image_output_tokens = image_part_count * 2_000_u64;
+    // Image output tokens per image vary by resolution, billed at $120/1M tokens
+    let tokens_per_image: u64 = match image_size.as_str() {
+        "1K" => 500,
+        "2K" => 1_000,
+        _    => 2_000, // 4K default
+    };
+    let image_output_tokens = image_part_count * tokens_per_image;
     // Remaining candidates tokens are text/thinking, billed at $12/1M tokens
     let text_output_tokens = candidates_tokens.saturating_sub(image_output_tokens);
 

src-tauri/src/lib.rs

@@ -1,7 +1,7 @@
 mod gemini;
 mod storage;
 
-use gemini::{call_gemini, read_reference_image_base64};
+use gemini::{call_gemini, read_reference_image_base64, GeminiCallParams};
 use serde::Serialize;
 use storage::{
     delete_thread_from_disk, load_config_from_disk, load_threads_from_disk, save_config_to_disk,
@@ -46,16 +46,30 @@ async fn save_config(config: Config) -> Result<(), String> {
 }
 
 #[tauri::command]
+#[allow(clippy::too_many_arguments)]
 async fn send_message(
     api_key: String,
     mode: String,
+    aspect_ratio: Option<String>,
+    image_size: String,
     history: Vec<ThreadMessage>,
     user_text: Option<String>,
     user_image_base64: Option<String>,
     user_image_mime: Option<String>,
 ) -> Result<SendMessageResult, String> {
-    let (parts, cost_usd) =
-        call_gemini(api_key, mode, history, user_text, user_image_base64, user_image_mime).await?;
+    let (parts, cost_usd) = call_gemini(
+        api_key,
+        history,
+        GeminiCallParams {
+            mode,
+            aspect_ratio,
+            image_size,
+            user_text,
+            user_image_base64,
+            user_image_mime,
+        },
+    )
+    .await?;
     Ok(SendMessageResult { parts, cost_usd })
 }
 

src-tauri/src/storage.rs

@@ -25,11 +25,19 @@ pub struct ThreadMessage {
     pub cost_usd: Option<f64>,
 }
 
+fn default_image_size() -> String {
+    "4K".to_string()
+}
+
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct Thread {
     pub id: String,
     pub name: String,
     pub mode: String,
+    #[serde(rename = "aspectRatio", skip_serializing_if = "Option::is_none")]
+    pub aspect_ratio: Option<String>,
+    #[serde(rename = "imageSize", default = "default_image_size")]
+    pub image_size: String,
     pub messages: Vec<ThreadMessage>,
     #[serde(rename = "createdAt")]
     pub created_at: i64,

src-tauri/tauri.conf.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://schema.tauri.app/config/2",
   "productName": "tatsumi",
-  "version": "1.0.0",
+  "version": "1.2.0",
   "identifier": "com.naomi.tatsumi",
   "build": {
     "beforeDevCommand": "pnpm dev",

src/app.tsx

@@ -16,17 +16,127 @@ import { SettingsModal } from "./components/settingsModal.tsx";
 import { Sidebar } from "./components/sidebar.tsx";
 import { ThreadView } from "./components/threadView.tsx";
 import { WelcomeScreen } from "./components/welcomeScreen.tsx";
-import type { Config, Mode, PendingInput, Thread } from "./types/index.ts";
+import type {
+  AspectRatio,
+  Config,
+  ImageSize,
+  Mode,
+  PendingInput,
+  Thread,
+} from "./types/index.ts";
 
-const generateThreadName = (mode: Mode, text?: string): string => {
-  if (
-    mode === "replace"
-    || text === undefined
-    || text.trim().length === 0
-  ) {
-    return `Replace - ${new Date().toLocaleString()}`;
-  }
-  return text.trim().slice(0, 40);
+const adjectives = [
+  "Amber",
+  "Ancient",
+  "Arctic",
+  "Azure",
+  "Blazing",
+  "Celestial",
+  "Cobalt",
+  "Crimson",
+  "Crystal",
+  "Dark",
+  "Distant",
+  "Ember",
+  "Emerald",
+  "Eternal",
+  "Ethereal",
+  "Fleeting",
+  "Frosted",
+  "Gilded",
+  "Golden",
+  "Hidden",
+  "Hollow",
+  "Indigo",
+  "Ivory",
+  "Jade",
+  "Lunar",
+  "Midnight",
+  "Mystic",
+  "Neon",
+  "Obsidian",
+  "Onyx",
+  "Opal",
+  "Pale",
+  "Pearl",
+  "Radiant",
+  "Rusted",
+  "Sapphire",
+  "Scarlet",
+  "Shattered",
+  "Silent",
+  "Silver",
+  "Spectral",
+  "Twilight",
+  "Velvet",
+  "Verdant",
+  "Veiled",
+  "Violet",
+  "Wandering",
+  "Wild",
+  "Woven",
+  "Arcane",
+];
+
+const nouns = [
+  "Abyss",
+  "Archive",
+  "Aria",
+  "Bloom",
+  "Cascade",
+  "Chronicle",
+  "Cipher",
+  "Compass",
+  "Crown",
+  "Dawn",
+  "Dream",
+  "Echo",
+  "Elegy",
+  "Ember",
+  "Equinox",
+  "Flame",
+  "Fragment",
+  "Garden",
+  "Horizon",
+  "Labyrinth",
+  "Lantern",
+  "Mirage",
+  "Mist",
+  "Murmur",
+  "Nexus",
+  "Nocturne",
+  "Oracle",
+  "Phantom",
+  "Prism",
+  "Requiem",
+  "Reverie",
+  "Ruin",
+  "Shadow",
+  "Shard",
+  "Silence",
+  "Solstice",
+  "Sonata",
+  "Specter",
+  "Storm",
+  "Tempest",
+  "Throne",
+  "Tide",
+  "Veil",
+  "Vortex",
+  "Whisper",
+  "Zenith",
+  "Sigil",
+  "Glyph",
+  "Dusk",
+  "Omen",
+];
+
+const generateThreadName = (): string => {
+  const adjectiveIndex = Math.floor(Math.random() * adjectives.length);
+  const nounIndex = Math.floor(Math.random() * nouns.length);
+  const adjective = adjectives[adjectiveIndex] ?? "Arcane";
+  const noun = nouns[nounIndex] ?? "Reverie";
+  return `${adjective} ${noun}`;
 };
 
 const generateId = (): string => {
@@ -67,13 +177,7 @@ const resolveUpdatedThread = (updatedThread: Thread): Thread => {
     return updatedThread;
   }
 
-  const firstTextPart = firstMessage.parts.find((part) => {
-    return part.type === "text";
-  });
-  const name = generateThreadName(
-    updatedThread.mode,
-    firstTextPart?.text,
-  );
+  const name = generateThreadName();
   return { ...updatedThread, name };
 };
 
@@ -120,11 +224,18 @@ const app = (): JSX.Element => {
   });
 
   const handleNewThread = useCallback(
-    (mode: Mode): void => {
+    (
+      mode:        Mode,
+      aspectRatio: AspectRatio | undefined,
+      imageSize:   ImageSize,
+    ): void => {
       const now = Date.now();
       const createdThread: Thread = {
+        /* eslint-disable-next-line stylistic/no-extra-parens -- Required for conditional spread with exactOptionalPropertyTypes */
+        ...(aspectRatio !== undefined && { aspectRatio }),
         createdAt: now,
         id:        generateId(),
+        imageSize: imageSize,
         messages:  [],
         mode:      mode,
         name:      `New ${mode} thread`,

src/components/inputArea.tsx

@@ -19,7 +19,12 @@ import {
   type JSX,
   type KeyboardEvent,
 } from "react";
-import type { Mode, PendingInput } from "../types/index.ts";
+import type {
+  AspectRatio,
+  ImageSize,
+  Mode,
+  PendingInput,
+} from "../types/index.ts";
 
 const dropZoneBaseClass = [
   "border-2 border-dashed rounded-xl p-8",
@@ -113,14 +118,19 @@ const readClipboardImage = async(
   onFileRead(new File([ blob ], "clipboard.png", { type: imageType }));
 };
 
-const modeLabelText = (mode: Mode): string => {
+const modeLabelText = (
+  mode:        Mode,
+  aspectRatio: AspectRatio | undefined,
+  imageSize:   ImageSize,
+): string => {
   if (mode === "avatar") {
-    return "🟣 Avatar Mode (1:1)";
+    return `🟣 Avatar Mode · ${imageSize}`;
   }
   if (mode === "art") {
-    return "🩷 Art Mode (16:9)";
+    const ratio = aspectRatio ?? "16:9";
+    return `🩷 Art Mode (${ratio}) · ${imageSize}`;
   }
-  return "🔵 Replace Mode";
+  return `🔵 Replace Mode · ${imageSize}`;
 };
 
 type OnSendCallback = (
@@ -130,7 +140,9 @@ type OnSendCallback = (
 )=> void;
 
 interface InputAreaProperties {
+  readonly aspectRatio?:         AspectRatio;
   readonly hasMessages:          boolean;
+  readonly imageSize:            ImageSize;
   readonly initialImageBase64?:  string;
   readonly initialImageMime?:    string;
   readonly initialImagePreview?: string;
@@ -144,7 +156,9 @@ interface InputAreaProperties {
 /**
  * Renders the input area for composing and sending messages.
  * @param props - The component props.
+ * @param props.aspectRatio - The thread's aspect ratio setting (Art mode only).
  * @param props.hasMessages - Whether the thread already has messages (affects replace mode UI).
+ * @param props.imageSize - The thread's resolution setting.
  * @param props.initialImageBase64 - Initial base64 image data to pre-populate.
  * @param props.initialImageMime - Initial image MIME type to pre-populate.
  * @param props.initialImagePreview - Initial image preview URL to pre-populate.
@@ -156,7 +170,9 @@ interface InputAreaProperties {
  * @returns The JSX element.
  */
 const inputArea = ({
+  aspectRatio,
   hasMessages,
+  imageSize,
   initialImageBase64,
   initialImageMime,
   initialImagePreview,
@@ -256,7 +272,8 @@ const inputArea = ({
     if (isInitialReplace && imageBase64 === undefined) {
       return;
     }
-    if (!isInitialReplace && text.trim().length === 0) {
+    const hasContent = text.trim().length > 0 || imageBase64 !== undefined;
+    if (!isInitialReplace && !hasContent) {
       return;
     }
 
@@ -332,11 +349,14 @@ const inputArea = ({
       : dropZoneInactiveClass,
   ].join(" ");
 
+  const hasNoContent = text.trim().length === 0 && imageBase64 === undefined;
+  const isSendDisabled = isLoading || hasNoContent;
+
   return (
     <div className="border-t border-purple-900/30 p-4 bg-[#0f0a1a]">
       <div className="flex items-center gap-2 mb-2">
         <span className="text-xs text-gray-500 uppercase tracking-wider">
-          {modeLabelText(mode)}
+          {modeLabelText(mode, aspectRatio, imageSize)}
         </span>
       </div>
 
@@ -397,6 +417,15 @@ const inputArea = ({
             type="file"
           />
 
+          <textarea
+            className={textareaClass}
+            disabled={isLoading}
+            onChange={handleTextChange}
+            placeholder="Add notes to include with the image (optional)..."
+            rows={2}
+            value={text}
+          />
+
           <button
             className={replaceButtonClass}
             disabled={isLoading || imageBase64 === undefined}
@@ -414,41 +443,40 @@ const inputArea = ({
         </div>
 
         : <div className="flex flex-col gap-3">
-          {mode === "replace"
-            ? <div className="flex flex-col gap-2">
-              {imagePreview === undefined
-                ? <button
-                  className={pasteButtonClass}
-                  onClick={handlePasteButtonClick}
+          <div className="flex flex-col gap-2">
+            {imagePreview === undefined
+              ? <button
+                className={pasteButtonClass}
+                onClick={handlePasteButtonClick}
+                type="button"
+              >
+                {mode === "replace"
+                  ? "📋 Paste replacement image (optional)"
+                  : "📋 Paste image (optional)"}
+              </button>
+
+              : <div className="relative inline-block">
+                <img
+                  alt="Upload preview"
+                  className="max-h-32 rounded-lg border border-purple-700/40"
+                  src={imagePreview}
+                />
+                <button
+                  className={clearButtonClass}
+                  onClick={clearImage}
                   type="button"
                 >
-                  {"📋 Paste replacement image (optional)"}
+                  {"×"}
                 </button>
-
-                : <div className="relative inline-block">
-                  <img
-                    alt="Upload preview"
-                    className="max-h-32 rounded-lg border border-purple-700/40"
-                    src={imagePreview}
-                  />
-                  <button
-                    className={clearButtonClass}
-                    onClick={clearImage}
-                    type="button"
-                  >
-                    {"×"}
-                  </button>
-                </div>}
-              <input
-                accept="image/*"
-                className="hidden"
-                onChange={handleFileChange}
-                ref={fileInputReference}
-                type="file"
-              />
-            </div>
-
-            : null}
+              </div>}
+            <input
+              accept="image/*"
+              className="hidden"
+              onChange={handleFileChange}
+              ref={fileInputReference}
+              type="file"
+            />
+          </div>
 
           <div className="flex gap-3 items-end">
             <textarea
@@ -464,7 +492,7 @@ const inputArea = ({
             />
             <button
               className={sendButtonClass}
-              disabled={isLoading || text.trim().length === 0}
+              disabled={isSendDisabled}
               onClick={handleSend}
               type="button"
             >

src/components/newThreadModal.tsx

@@ -1,12 +1,18 @@
 /**
- * @file Modal for selecting a new thread generation mode.
+ * @file Modal for selecting a new thread generation mode and options.
  * @copyright Naomi Carrigan
  * @license Naomi's Public License
  * @author Naomi Carrigan
  */
 /* eslint-disable max-lines-per-function -- Component JSX inherently requires many lines */
-import { useCallback, type JSX, type MouseEvent } from "react";
-import type { Mode } from "../types/index.ts";
+/* eslint-disable max-lines -- Two-step modal requires many option definitions */
+import {
+  useCallback,
+  useState,
+  type JSX,
+  type MouseEvent,
+} from "react";
+import type { AspectRatio, ImageSize, Mode } from "../types/index.ts";
 
 interface ModeOption {
   colour:      string;
@@ -16,14 +22,25 @@ interface ModeOption {
   mode:        Mode;
 }
 
+interface AspectRatioOption {
+  label: string;
+  value: AspectRatio;
+}
+
+interface ImageSizeOption {
+  description: string;
+  label:       string;
+  value:       ImageSize;
+}
+
 const avatarDescription = [
   "Generate a square 1:1 portrait.",
   "Perfect for profile pictures and avatars.",
 ].join(" ");
 
 const artDescription = [
-  "Generate wide 16:9 landscape artwork.",
-  "Great for wallpapers and banners.",
+  "Generate landscape or portrait artwork.",
+  "Choose your aspect ratio and resolution.",
 ].join(" ");
 
 const replaceDescription = [
@@ -55,6 +72,21 @@ const modeOptionList: Array<ModeOption> = [
   },
 ];
 
+const aspectRatioOptions: Array<AspectRatioOption> = [
+  { label: "1:1 Square", value: "1:1" },
+  { label: "4:3 Standard", value: "4:3" },
+  { label: "3:4 Portrait", value: "3:4" },
+  { label: "16:9 Widescreen", value: "16:9" },
+  { label: "9:16 Wallpaper", value: "9:16" },
+  { label: "21:9 Ultrawide", value: "21:9" },
+];
+
+const imageSizeOptions: Array<ImageSizeOption> = [
+  { description: "Faster, cheaper", label: "1K", value: "1K" },
+  { description: "Balanced", label: "2K", value: "2K" },
+  { description: "Best quality", label: "4K", value: "4K" },
+];
+
 const colourMap: Record<
   string,
   { badge: string; button: string; hover: string }
@@ -82,6 +114,20 @@ const isMode = (value: string): value is Mode => {
   return validModesSet.has(value);
 };
 
+const validAspectRatios = new Set<string>([
+  "1:1", "3:4", "4:3", "9:16", "16:9", "21:9",
+]);
+
+const isAspectRatio = (value: string): value is AspectRatio => {
+  return validAspectRatios.has(value);
+};
+
+const validImageSizes = new Set<string>([ "1K", "2K", "4K" ]);
+
+const isImageSize = (value: string): value is ImageSize => {
+  return validImageSizes.has(value);
+};
+
 const overlayClass = [
   "fixed inset-0 bg-black/70 backdrop-blur-sm",
   "flex items-center justify-center z-50",
@@ -93,32 +139,243 @@ const panelClass = [
   "shadow-2xl shadow-purple-900/30",
 ].join(" ");
 
+const pillBaseClass = [
+  "px-4 py-2 rounded-xl border text-sm font-medium",
+  "transition-all duration-200 cursor-pointer",
+].join(" ");
+
+const pillActiveClass = "border-purple-500 bg-purple-600/30 text-white";
+
+const pillInactiveClass = [
+  "border-purple-900/40 bg-transparent text-gray-400",
+  "hover:border-purple-500/60 hover:text-gray-200",
+].join(" ");
+
+const startButtonClass = [
+  "w-full bg-gradient-to-r from-purple-600 to-pink-600",
+  "hover:from-purple-500 hover:to-pink-500",
+  "text-white font-semibold py-3 rounded-xl",
+  "transition-all duration-200 mt-2",
+].join(" ");
+
+const sizeButtonActiveClass = "border-purple-500 bg-purple-600/30 text-white";
+
+const sizeButtonInactiveClass = [
+  "border-purple-900/40 bg-transparent text-gray-400",
+  "hover:border-purple-500/60 hover:text-gray-200",
+].join(" ");
+
+const backButtonClass = [
+  "text-gray-400 hover:text-white",
+  "transition-colors text-xl",
+].join(" ");
+
+const sectionLabelClass = [
+  "text-sm text-gray-400",
+  "uppercase tracking-wider mb-3",
+].join(" ");
+
 interface NewThreadModalProperties {
   readonly onClose:  ()=> void;
-  readonly onSelect: (mode: Mode)=> void;
+  readonly onSelect: (
+    mode:        Mode,
+    aspectRatio: AspectRatio | undefined,
+    imageSize:   ImageSize,
+  )=> void;
 }
 
 /**
- * Renders the new thread modal for selecting a generation mode.
+ * Renders the new thread modal for selecting a generation mode and options.
  * @param props - The component props.
  * @param props.onClose - Callback to close the modal.
- * @param props.onSelect - Callback when a mode is selected.
+ * @param props.onSelect - Callback when a mode and options are confirmed.
  * @returns The JSX element.
  */
 const threadModal = ({
   onClose,
   onSelect,
 }: NewThreadModalProperties): JSX.Element => {
+  const [ step, setStep ] = useState<"mode" | "options">("mode");
+  const [ selectedMode, setSelectedMode ] = useState<Mode | undefined>(
+    undefined,
+  );
+  const [ aspectRatio, setAspectRatio ] = useState<AspectRatio>("16:9");
+  const [ imageSize, setImageSize ] = useState<ImageSize>("4K");
+
   const handleModeSelect = useCallback(
     (event: MouseEvent<HTMLButtonElement>): void => {
       const rawMode = event.currentTarget.dataset.mode;
       if (rawMode !== undefined && isMode(rawMode)) {
-        onSelect(rawMode);
+        setSelectedMode(rawMode);
+        setStep("options");
       }
     },
-    [ onSelect ],
+    [],
   );
 
+  const handleBack = useCallback((): void => {
+    setStep("mode");
+  }, []);
+
+  const handleStart = useCallback((): void => {
+    if (selectedMode === undefined) {
+      return;
+    }
+    const ratio = selectedMode === "art"
+      ? aspectRatio
+      : undefined;
+    onSelect(selectedMode, ratio, imageSize);
+  }, [ selectedMode, aspectRatio, imageSize, onSelect ]);
+
+  const handleAspectRatioSelect = useCallback(
+    (event: MouseEvent<HTMLButtonElement>): void => {
+      const rawRatio = event.currentTarget.dataset.ratio;
+      if (rawRatio !== undefined && isAspectRatio(rawRatio)) {
+        setAspectRatio(rawRatio);
+      }
+    },
+    [],
+  );
+
+  const handleImageSizeSelect = useCallback(
+    (event: MouseEvent<HTMLButtonElement>): void => {
+      const rawSize = event.currentTarget.dataset.size;
+      if (rawSize !== undefined && isImageSize(rawSize)) {
+        setImageSize(rawSize);
+      }
+    },
+    [],
+  );
+
+  const selectedModeOption = modeOptionList.find((o) => {
+    return o.mode === selectedMode;
+  });
+  const selectedColours = selectedModeOption === undefined
+    ? colourMap.purple
+    : colourMap[selectedModeOption.colour];
+
+  if (step === "options" && selectedModeOption !== undefined) {
+    return (
+      <div className={overlayClass}>
+        <div className={panelClass}>
+          <div className="flex items-center gap-3 mb-6">
+            <button
+              className={backButtonClass}
+              onClick={handleBack}
+              type="button"
+            >
+              {"←"}
+            </button>
+            <h2 className="text-2xl font-bold text-white flex-1">
+              {"Configure Thread"}
+            </h2>
+            <button
+              className="text-gray-400 hover:text-white transition-colors"
+              onClick={onClose}
+              type="button"
+            >
+              {"×"}
+            </button>
+          </div>
+
+          <div
+            className={[
+              "flex items-center gap-3 mb-6 p-4",
+              "rounded-xl bg-[#241836] border border-purple-900/30",
+            ].join(" ")}
+          >
+            <span className="text-3xl">{selectedModeOption.icon}</span>
+            <div>
+              <div className="flex items-center gap-2">
+                <span className="text-white font-semibold">
+                  {selectedModeOption.label}
+                </span>
+                <span
+                  className={`text-xs px-2 py-0.5 rounded-full ${selectedColours?.badge ?? ""}`}
+                >
+                  {selectedModeOption.mode}
+                </span>
+              </div>
+              <p className="text-gray-400 text-xs mt-0.5">
+                {selectedModeOption.description}
+              </p>
+            </div>
+          </div>
+
+          {selectedMode === "art"
+            ? <div className="mb-6">
+              <p className={sectionLabelClass}>
+                {"Aspect Ratio"}
+              </p>
+              <div className="flex flex-wrap gap-2">
+                {aspectRatioOptions.map((option) => {
+                  const pillClass = [
+                    pillBaseClass,
+                    aspectRatio === option.value
+                      ? pillActiveClass
+                      : pillInactiveClass,
+                  ].join(" ");
+                  return (
+                    <button
+                      className={pillClass}
+                      data-ratio={option.value}
+                      key={option.value}
+                      onClick={handleAspectRatioSelect}
+                      type="button"
+                    >
+                      {option.label}
+                    </button>
+                  );
+                })}
+              </div>
+            </div>
+
+            : null}
+
+          <div className="mb-6">
+            <p className={sectionLabelClass}>
+              {"Resolution"}
+            </p>
+            <div className="flex gap-3">
+              {imageSizeOptions.map((option) => {
+                const sizeClass = [
+                  "flex-1 flex flex-col items-center py-3 px-2",
+                  "rounded-xl border text-sm font-medium",
+                  "transition-all duration-200 cursor-pointer",
+                  imageSize === option.value
+                    ? sizeButtonActiveClass
+                    : sizeButtonInactiveClass,
+                ].join(" ");
+                return (
+                  <button
+                    className={sizeClass}
+                    data-size={option.value}
+                    key={option.value}
+                    onClick={handleImageSizeSelect}
+                    type="button"
+                  >
+                    <span className="text-base font-bold">{option.label}</span>
+                    <span className="text-xs mt-0.5 opacity-70">
+                      {option.description}
+                    </span>
+                  </button>
+                );
+              })}
+            </div>
+          </div>
+
+          <button
+            className={startButtonClass}
+            onClick={handleStart}
+            type="button"
+          >
+            {"Start Thread →"}
+          </button>
+        </div>
+      </div>
+    );
+  }
+
   return (
     <div className={overlayClass}>
       <div className={panelClass}>

src/components/threadView.tsx

@@ -68,19 +68,19 @@ const modeLabel = (mode: Mode): string => {
 interface BuildUserPartsOptions {
   readonly imageBase64?: string;
   readonly imageMime?:   string;
-  readonly mode:         Mode;
   readonly text:         string;
 }
 
 const buildUserParts = ({
   imageBase64,
   imageMime,
-  mode,
   text,
 }: BuildUserPartsOptions): Array<MessagePart> => {
   const userParts: Array<MessagePart> = [];
 
-  if (mode === "replace" && imageBase64 !== undefined) {
+  if (imageBase64 === undefined) {
+    userParts.push({ text: text, type: "text" });
+  } else {
     userParts.push({
       imageData: imageBase64,
       mimeType:  imageMime ?? "image/png",
@@ -89,8 +89,6 @@ const buildUserParts = ({
     if (text.length > 0) {
       userParts.push({ text: text, type: "text" });
     }
-  } else {
-    userParts.push({ text: text, type: "text" });
   }
 
   return userParts;
@@ -174,13 +172,18 @@ const threadView = ({
       onLoadingChange(true);
       onErrorChange(undefined);
 
-      const { messages, mode } = thread;
+      const {
+        aspectRatio,
+        imageSize: rawImageSize,
+        messages,
+        mode,
+      } = thread;
+      const imageSize = rawImageSize ?? "4K";
       const userParts = buildUserParts({
         /* eslint-disable-next-line stylistic/no-extra-parens -- Required for conditional spread with exactOptionalPropertyTypes */
         ...(imageBase64 !== undefined && { imageBase64 }),
         /* eslint-disable-next-line stylistic/no-extra-parens -- Required for conditional spread with exactOptionalPropertyTypes */
         ...(imageMime !== undefined && { imageMime }),
-        mode,
         text,
       });
 
@@ -217,7 +220,9 @@ const threadView = ({
             "send_message",
             {
               apiKey,
+              aspectRatio,
               history,
+              imageSize,
               mode,
               userImageBase64,
               userImageMime,
@@ -253,7 +258,13 @@ const threadView = ({
 
   const handleRetry = useCallback(
     (modelMessageIndex: number): void => {
-      const { messages, mode } = thread;
+      const {
+        aspectRatio,
+        imageSize: rawImageSize,
+        messages,
+        mode,
+      } = thread;
+      const imageSize = rawImageSize ?? "4K";
       const userMessageIndex = modelMessageIndex - 1;
       const userMessage = messages[userMessageIndex];
 
@@ -300,7 +311,9 @@ const threadView = ({
           }
           const response = await invoke<SendResult>("send_message", {
             apiKey,
+            aspectRatio,
             history,
+            imageSize,
             mode,
             userImageBase64,
             userImageMime,
@@ -353,7 +366,13 @@ const threadView = ({
 
   const handleEditCommit = useCallback(
     (messageIndex: number, editedText: string): void => {
-      const { messages, mode } = thread;
+      const {
+        aspectRatio,
+        imageSize: rawImageSize,
+        messages,
+        mode,
+      } = thread;
+      const imageSize = rawImageSize ?? "4K";
       const originalMessage = messages[messageIndex];
       if (originalMessage === undefined) {
         return;
@@ -404,7 +423,9 @@ const threadView = ({
           }
           const response = await invoke<SendResult>("send_message", {
             apiKey,
+            aspectRatio,
             history,
+            imageSize,
             mode,
             userImageBase64,
             userImageMime,
@@ -536,7 +557,11 @@ const threadView = ({
         {...(pendingInput?.text !== undefined && {
           initialText: pendingInput.text,
         })}
+        {...(thread.aspectRatio !== undefined && {
+          aspectRatio: thread.aspectRatio,
+        })}
         hasMessages={thread.messages.length > 0}
+        imageSize={thread.imageSize ?? "4K"}
         isLoading={isLoading}
         mode={thread.mode}
         onInputChange={onPendingInputChange}

src/types/index.ts

@@ -5,6 +5,8 @@
  * @author Naomi Carrigan
  */
 
+type AspectRatio = "1:1" | "3:4" | "4:3" | "9:16" | "16:9" | "21:9";
+type ImageSize = "1K" | "2K" | "4K";
 type Mode = "avatar" | "art" | "replace";
 
 interface Config {
@@ -33,12 +35,23 @@ interface ThreadMessage {
 }
 
 interface Thread {
-  createdAt: number;
-  id:        string;
-  messages:  Array<ThreadMessage>;
-  mode:      Mode;
-  name:      string;
-  updatedAt: number;
+  aspectRatio?: AspectRatio;
+  createdAt:    number;
+  id:           string;
+  imageSize?:   ImageSize;
+  messages:     Array<ThreadMessage>;
+  mode:         Mode;
+  name:         string;
+  updatedAt:    number;
 }
 
-export type { Config, MessagePart, Mode, PendingInput, Thread, ThreadMessage };
+export type {
+  AspectRatio,
+  Config,
+  ImageSize,
+  MessagePart,
+  Mode,
+  PendingInput,
+  Thread,
+  ThreadMessage,
+};