chore(model): add fields for image2image input

instill-ai · Dec 3, 2023 · 5a963a8 · 5a963a8
1 parent 5a27bfd
commit 5a963a8
Show file tree

Hide file tree

Showing 2 changed files with 38 additions and 15 deletions.
diff --git a/model/model/v1alpha/task_image_to_image.proto b/model/model/v1alpha/task_image_to_image.proto
@@ -8,17 +8,25 @@ import "model/model/v1alpha/common.proto";
 
 // ImageToImageInput represents the input of image to image task
 message ImageToImageInput {
-  // Input Image
+  // The Prompt Image, only for multimodal input
   oneof type {
     // Image type URL
-    string image_url = 1;
+    string prompt_image_url = 1;
     // Image type base64
-    string image_base64 = 2;
+    string prompt_image_base64 = 2;
   }
-  // The text input of the model
-  optional string prompt = 3 [(google.api.field_behavior) = OPTIONAL];
-  // Additional parameters of the model
-  repeated ExtraParamObject extra_params = 4 [(google.api.field_behavior) = OPTIONAL];
+  // The prompt text
+  optional string prompt = 3 [(google.api.field_behavior) = REQUIRED];
+  // The steps, default is 5
+  optional int32 steps = 4 [(google.api.field_behavior) = OPTIONAL];
+  // The guidance scale, default is 7.5
+  optional float cfg_scale = 5 [(google.api.field_behavior) = OPTIONAL];
+  // The seed, default is 0
+  optional int32 seed = 6 [(google.api.field_behavior) = OPTIONAL];
+  // The number of generated samples, default is 1
+  optional int32 samples = 7 [(google.api.field_behavior) = OPTIONAL];
+  // The extra parameters
+  optional ExtraParamObject extra_params = 8 [(google.api.field_behavior) = OPTIONAL];
 }
 
 // ImageToImageOutput represents the output of image to image task

diff --git a/openapiv2/openapiv2.swagger.yaml b/openapiv2/openapiv2.swagger.yaml
@@ -8218,22 +8218,37 @@ definitions:
   v1alphaImageToImageInput:
     type: object
     properties:
-      image_url:
+      prompt_image_url:
         type: string
         title: Image type URL
-      image_base64:
+      prompt_image_base64:
         type: string
         title: Image type base64
       prompt:
         type: string
-        title: The text input of the model
+        title: The prompt text
+      steps:
+        type: integer
+        format: int32
+        title: The steps, default is 5
+      cfg_scale:
+        type: number
+        format: float
+        title: The guidance scale, default is 7.5
+      seed:
+        type: integer
+        format: int32
+        title: The seed, default is 0
+      samples:
+        type: integer
+        format: int32
+        title: The number of generated samples, default is 1
       extra_params:
-        type: array
-        items:
-          type: object
-          $ref: '#/definitions/v1alphaExtraParamObject'
-        title: Additional parameters of the model
+        $ref: '#/definitions/v1alphaExtraParamObject'
+        title: The extra parameters
     title: ImageToImageInput represents the input of image to image task
+    required:
+      - prompt
   v1alphaImageToImageOutput:
     type: object
     properties: