vision: real camera frames to the model (image_url), auto-detected for multimodal ollama models; --vision/--no-vision flags; verified gemma4:12b sees frames and completed the beacon mission

testbed/README.md

@@ -101,6 +101,20 @@ works: `python -m testbed.chat --base-url https://api.openai.com/v1
 --model gpt-4o-mini --api-key $OPENAI_API_KEY`. `--auto-steps N` controls how
 many tool steps the model may chain per request (0 = one action per turn).
 
+### Real vision
+
+By default the model gets frames as a color-grid digest (works for any
+text-only model). If the model can actually see images (gemma3/gemma4,
+qwen2.5-vl, gpt-4o-mini, ...), pass the frames as real images:
+
+- auto-detected for local ollama (`/api/show` capabilities) — nothing to do
+- otherwise force it: `--vision` (or `--no-vision` to disable)
+
+With vision ON, `vision` tool results attach the actual camera frame as an
+image to the conversation — the model sees the crate, the wall, the glowing
+beacon, and orients itself. Verified locally: gemma4:12b completed a beacon
+mission on its first attempt using look_at/move navigation.
+
 ## Real-time mode
 
 Watch the capsule drive live in your browser: