renderdoc-inspect

RenderDoc Inspect

These tools replay an existing .rdc through RenderDoc's own replay API, so they expose what the RenderDoc UI can show rather than only what renderdoccmd can serialise. Every tool takes capture_file and never modifies it; only get_texture_data and get_buffer_data write to disk, and only to a destination you name.

Backends and capability gating

Two backends provide RenderDoc capabilities:

• Baseline — renderdoccmd, headless and always preferred for capture, conversion, thumbnails, and the remote server. It can only replay a capture, never read data out of one.
• Deep replay — renderdoc.pyd, which carries actions, resources, pipeline state, shaders, and readback. It is not importable by the adapter's own interpreter, so this skill reaches it through the Python interpreter bundled with qrenderdoc; qrenderdoc must therefore sit beside renderdoccmd.

replay_capabilities reports which backends are reachable and which capability groups they unlock (inspect, debug, perf, ext). When the deep backend is missing, the deep tools fail with one explicit message naming the missing backend and how to enable it, instead of crashing.

Working order

1. replay_capabilities — confirm the deep backend is reachable.
2. describe_capture — capture size, resource counts, and per-capture replay flags.
3. list_actions — find an event ID; name_filter and flag_filter do the searching.
4. get_action / get_pipeline_state — what one event is and what it bound.
5. get_shader_info — reflection, source, disassembly, and constant buffer values.
6. get_resource_usage, get_texture_data, get_buffer_data — the data behind a resource.

Replay runs in the Python interpreter bundled with qrenderdoc. On headless Linux, run the adapter under Xvfb or provide another working X/Wayland display; the official archive does not include Qt's offscreen platform plugin.

Large results are paged with offset and limit; keep limit small enough to read.