@lritter Maybe, but then you might start bringing in some shadow artifacts. It's worth testing if there's a sweet spot.
Owner/Founder Burning Candle Software (available for contract/freelance work); Graphics, distributed systems, computational geometry, machine learning, optimization and compression.
@lritter Maybe, but then you might start bringing in some shadow artifacts. It's worth testing if there's a sweet spot.
@lritter Sure, from my experiments with similar things in the past, things like a sweeping line do turn into a ghosted smear... but it would look good when the viewer was still.
Part of the problem is that because the line moves with the viewer, on roughly flatish surfaces it is static on the screen, so it is visible when movement happens. But you end up with a smear instead of a hard boundary.
@lritter You wouldn't need the stochasticness in that case though.
@lritter This is why I was saying clip maps and tiles would work, because you'd transition into the new tile and that would blend into the old lower or higher resolution one from the previous frame...
There could be other benefits like lazy updates and sparseness too.
@lritter bingo!
@lritter the cascade boundary isn't spatially stable vs the scene though! hence the ghosting
@lritter sure, but the shadows themselves are stable spatially for static geometry (you end up with some ghosting where a dynamic object moves but the object its shadow falls on is static).
@lritter I think it's probably cos the cascade boundary moves frame to frame with the viewer, so you'd probably end up with some weird ghosting in motion.
There might be some solutions using lazily updated tiles, more like a clip map.
@shapr Normalise saying no to bison, even if they are cheap.
@shapr If you have to ask where to store a bison, you should probably not be buying a bison.
@lritter imagine the blooper reel
@dotstdy @pervognsen Life is so much simpler with unified memory and an RTOS.
@pervognsen could take the approach used in embedded (drawing or compositing latency sensitive stuff on an interrupt) and extend it to the GPU world by late latching just latency sensitive things (in focus text, scrolling and dragging). Seems a little bit at odds with the imgui approach though.
@aras @rygorous There's an interesting take on this from Yanis Yaroufakis, that we are moving into a techo-feudalist society where people till the virtual soil that is owned by a digital lord and that what we call "content creation" is largely that, a narrow form of creation at the behest of the people that own the infrastructure that delivers it.
There is some ideological tilt there, but I think in that framing the ideas aren't that opposing, just different sides of the same phenomenon.
@simonf @neilhenning One of the adjustments is the amount of extra context in your head. Becoming a dad turned me from a "I can carry all I need to do in my head" guy into a "hierarchal dot point mind map" guy to externalise mental context to keep on top of everything.
@rygorous that's because the singular is Fergi, one I multiple Us.
Found this cool sculpture in Sacramento.
We live in an age of miracle and wonder, but it's mostly owned by someone else with questionable judgment.
Don't shit bricks if you live in a ass house
Back in the day, Twitter used to be really good if I set up a thread to connect freelancers with clients. Let's see if Mastodon can do it.
Clients: if you're looking for freelancers/contractors, get in the comments
Freelancers/contractors: get in the comments
Everyone else: boosts appreciated.
The market is *dead* for freelancers and a big part of that (in my opinion) is fragmentation. Let's get that network effect *back*.