diff --git a/js/msr_character.js b/js/msr_character.js
index 957c075..b9f7fc7 100644
--- a/js/msr_character.js
+++ b/js/msr_character.js
@@ -1,5 +1,6 @@
 const { app } = window.comfyAPI.app;
 const { api } = window.comfyAPI.api;
+const DEFAULT_ANALYZE_PROMPT = "Describe the character's physical appearance in two concise sentences. Specify their hair color/style, face details, and their clothing type/color. Keep the entire response very brief.";
 
 function findWidget(node, name) {
   return (node.widgets || []).find((widget) => widget.name === name);
@@ -369,15 +370,19 @@ function buildCharacterUi(node) {
     const providerWidget = findWidget(node, "analyze_provider");
     const baseUrlWidget = findWidget(node, "analyze_base_url");
     const modelWidget = findWidget(node, "analyze_model");
+    const promptWidget = findWidget(node, "analyze_prompt");
     const provider = window.prompt("Analyze provider: ollama, lmstudio, custom, off", providerWidget?.value || "ollama");
     if (provider == null) return;
     const baseUrl = window.prompt("Analyze base URL (blank = default)", baseUrlWidget?.value || "");
     if (baseUrl == null) return;
     const model = window.prompt("Analyze model (blank = provider default)", modelWidget?.value || "");
     if (model == null) return;
+    const prompt = window.prompt("Analyze prompt (blank = default)", promptWidget?.value || DEFAULT_ANALYZE_PROMPT);
+    if (prompt == null) return;
     setWidgetValue(node, providerWidget, provider.trim() || "ollama");
     setWidgetValue(node, baseUrlWidget, baseUrl.trim());
     setWidgetValue(node, modelWidget, model.trim());
+    setWidgetValue(node, promptWidget, prompt.trim());
     syncFormFromWidgets(node);
     node.setDirtyCanvas?.(true, true);
   });
@@ -461,6 +466,7 @@ async function analyzeCharacterNode(node, buttonWidget) {
   const providerWidget = findWidget(node, "analyze_provider");
   const baseUrlWidget = findWidget(node, "analyze_base_url");
   const modelWidget = findWidget(node, "analyze_model");
+  const promptWidget = findWidget(node, "analyze_prompt");
 
   buttonWidget.label = "Analyzing...";
   if (node._msrAnalyzeButton) {
@@ -489,6 +495,7 @@ async function analyzeCharacterNode(node, buttonWidget) {
         provider: providerWidget?.value || "ollama",
         base_url: baseUrlWidget?.value || "",
         model: modelWidget?.value || "",
+        prompt: promptWidget?.value || "",
         image_b64: images,
         image_debug: imageDebug,
       }),
@@ -560,6 +567,9 @@ app.registerExtension({
         if (!findWidget(this, "analyze_model")) {
           this.addWidget("text", "analyze_model", "");
         }
+        if (!findWidget(this, "analyze_prompt")) {
+          this.addWidget("text", "analyze_prompt", DEFAULT_ANALYZE_PROMPT);
+        }
 
         if (!this._msrPreviewWidget) {
           const previewContainer = buildCharacterUi(this);
@@ -587,7 +597,7 @@ app.registerExtension({
           return result;
         };
 
-        ["alias", "description", "analyze_provider", "analyze_base_url", "analyze_model"].forEach((widgetName) => {
+        ["alias", "description", "analyze_provider", "analyze_base_url", "analyze_model", "analyze_prompt"].forEach((widgetName) => {
           hideNodeWidget(findWidget(this, widgetName));
         });
         hideNodeWidget((this.widgets || []).find((widget) => widget.msrAnalyze));
diff --git a/ltx_director.py b/ltx_director.py
index d3e48e5..48fcbec 100644
--- a/ltx_director.py
+++ b/ltx_director.py
@@ -512,6 +512,11 @@ _ANALYZE_PROMPT = (
 )
 
 
+def _resolve_analyze_prompt(data: dict) -> str:
+    prompt = (data.get("prompt") or "").strip()
+    return prompt or _ANALYZE_PROMPT
+
+
 def _extract_ollama_generated_text(resp_json: dict) -> str:
     if not isinstance(resp_json, dict):
         return ""
@@ -621,6 +626,7 @@ async def analyze_character_endpoint(request):
         image_debug = data.get("image_debug") or []
         char_index = int(data.get("char_index", 0))
         provider, base_url, model_name = _resolve_provider(data)
+        analyze_prompt = _resolve_analyze_prompt(data)
 
         if provider == "off":
             return web.json_response({"status": "error", "message": "Analyze is set to Off / Manual."})
@@ -652,7 +658,7 @@ async def analyze_character_endpoint(request):
                     payload = {
                         "model": model_name,
                         "system": _ANALYZE_SYSTEM_PROMPT,
-                        "prompt": _ANALYZE_PROMPT,
+                        "prompt": analyze_prompt,
                         "images": analysis_images,
                         "stream": False,
                         "keep_alive": 0,
@@ -691,7 +697,7 @@ async def analyze_character_endpoint(request):
                                 },
                                 {
                                     "role": "user",
-                                    "content": _ANALYZE_PROMPT,
+                                    "content": analyze_prompt,
                                     "images": analysis_images,
                                 },
                             ],
@@ -720,7 +726,7 @@ async def analyze_character_endpoint(request):
                         })
                 else:
                     # OpenAI-compatible vision chat (LM Studio / Custom).
-                    content = [{"type": "text", "text": _ANALYZE_PROMPT}]
+                    content = [{"type": "text", "text": analyze_prompt}]
                     for b64 in cleaned_b64_list:
                         content.append({"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{b64}"}})
                     payload = {