@99percentpeople/pi-codex-api 0.2.2 → 0.2.5
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +32 -0
- package/dist/index.ts +553 -49
- package/dist/index.ts.map +6 -6
- package/package.json +1 -1
- package/skills/gpt-image-prompts/SKILL.md +16 -0
package/package.json
CHANGED
|
@@ -133,6 +133,22 @@ When composition matters, specify:
|
|
|
133
133
|
|
|
134
134
|
Use left/right placement only when the user or intended layout requires it.
|
|
135
135
|
|
|
136
|
+
### Aspect ratio control
|
|
137
|
+
|
|
138
|
+
Express the target aspect ratio with explicit composition words in the prompt,
|
|
139
|
+
not with parameters:
|
|
140
|
+
|
|
141
|
+
- `Wide 16:9 landscape composition, panoramic` for a landscape canvas;
|
|
142
|
+
- `Tall 9:16 vertical portrait composition` for a tall vertical canvas;
|
|
143
|
+
- `Square 1:1 composition` for a square canvas;
|
|
144
|
+
- `Vertical 4:5 portrait composition` for a moderate portrait canvas.
|
|
145
|
+
|
|
146
|
+
GPT Image 2 honors these phrases and returns output that matches the stated
|
|
147
|
+
ratio. Sizing or aspect parameters exposed by the calling tool are often
|
|
148
|
+
ignored by the backend, so the prompt is the reliable control. Mention the
|
|
149
|
+
orientation early (purpose or composition line) rather than burying it at the
|
|
150
|
+
end.
|
|
151
|
+
|
|
136
152
|
## Text inside images
|
|
137
153
|
|
|
138
154
|
Text is a visual constraint, not an afterthought:
|