OpenDLSS lacks essential inputs, causing high cost and unclear training data
Problem
OpenDLSS does not accept the z‑buffer or other useful control signals, resulting in incomplete control; single‑frame processing at 1080p takes about 8 ms, inflating the rendering budget; users question the usefulness of the approach without per‑game trained weights or models, relying only on a generalized model.
Pain points
- Missing z‑buffer input results in incomplete control
- Single‑frame processing at 1080p takes about 8 ms, inflating rendering budget
- No per‑game trained weights; only a generalized model is available
Questions to verify
- Can equivalent control quality be achieved without providing the z‑buffer?
- Is the 8 ms latency of this implementation acceptable in real‑world gaming scenarios?
- Is there a plan to provide per‑game specific trained models or weights?
Evidence
quote · @Anonymous
I'm rather surprised that it doesn't take the z-buffer as an input. I would have thought that would have provided useful information, it's one of the more useful forms of contolnet.
quote · @Anonymous
Almost 8ms on 1080p resolution seems extremely expensive, does the original also eat into the rendering budget as much?
quote · @Anonymous
How useful is this without weights?Isn't the mote that Nvidia has is they work with studios to generate the training data from the game, then they ship a model per game?Or is my knowledge outdated here and they're just using a single generalised model?
Disclaimer: this card is distilled by AI from community discussions as “pain point evidence”. It is a starting point for need analysis and validation, not a “validated market need”. AI has analytical limits; please open the original discussion above to manually verify before key decisions.