- User Since
- Jan 13 2019, 6:00 AM (26 w, 6 d)
Thu, Jul 18
Mon, Jul 15
Actually, the delete bakes does work, I was mistakingly expecting it to clear the unbaked cache. Is there a way to do this other than changing a setting? For example if I move a helper that's driving a cloth sim, I need to recalculate the cache, but it doesnt seem possible unless I change a setting in the cloth.
Sun, Jul 14
This is fixed now in 2.8 latest build.
Hi Philipp, i've figured out the alt key issue, it seems to be a designed behaviour. If you press the alt key before clicking the box, then the values all change to be the new value, but if you press alt whilst pressing enter then the value is changed relative to each objects current value. So for example if they're all set to 0, and you then change one of them to 0.5, then change it to 0.6, but this time pressing alt when pressing enter, then it will change to 0.6 but the others will change to 0.1 (increasing by the same amount as the active object).
Sat, Jul 13
steps _ run sim, stop sim before end, press delete bakes. Please watch the video.
Fri, Jul 12
Just tested the release candidate and the bug is still here. Strangely it did appear to work once, with strange results, but after that it went back to only taking into account frame one of the rigid body simulation.
Mon, Jul 8
Thanks Brecht. Is there an alternative non properties panel related way of achieving the same in edit mode? I thought ctrl + L might be available to copy the materials from the active face to the other selected faces (like in object mode), but that brings up a select linked all menu instead.
Sun, Jul 7
same proble here, is there a solution yet? I'm using todays 2.8 build.
Sat, Jul 6
Fri, Jul 5
Wed, Jun 26
The file you sent does the same thing, you just forgot to put dof on. Here's your file with dof on.
Hi Sebastian, here's a video showing the issue. The problem occurs only when hashed blend alpha mode is enabled on the material settings and the principled shader has alpha set to less than 100%. Changing the dof distance changes where the banding is. The issue happens with or without an environment light.
Jun 19 2019
Jun 13 2019
Any ETA on someone looking at this, it's driving me nuts :)
Jun 4 2019
Thanks Brecht. Does that mean it's not possible to use the compositor to output individual frames to files correctly at render time if filmic is turned on? Basically all passes (including normal/alpha etc) will have filmic applied to them?
May 16 2019
May 14 2019
Doh, ok thanks. When will the last daily be available from the official download link rather than the experimental 'do not use in production' link? Really want those updates, but my client won't update their farm with anything but the official release.
May 13 2019
Thanks Brecht, are there any backwards compatibility issues with the 2.79b daily builds and the 2.79b official release? I'll probably update if there aren't any issues (just being careful because my client won't upgrade until the daily build replaces the official build)
display scaling is 150% on the 4k monitor and 100% on the 1080p display. Changing to 100% on the 4k display still shows the dot version of the crosshair instead of the solid white lines crosshair.
Here's a video showing the problem
Hi Christopher, it's a little better, but I think the problem is the crosshair style. The dots which make the cursor up are still barely visible, particularly when there's a lot of geometry. Just to clarify, if I drag blender over to my 1080p monitor, then the cursor returns to being a solid white crosshair and is easily visible, but on the 4k screen, it turns to the barely visible crosshair which is made of 8 small dots rather than being solid lines.
May 12 2019
Hi Marco, screencast replaces the crosshair shown in blender with a much thicker solid white crosshair which is easy to see. In blender the crosshair appears as 4 small outer dots (one at each end of the cross), and 4 small dots in the centre of the crosshair. The outer dots are approximately 2 pixels wide, and the centre dots are approximately 1 pixel wide, which as I'm sure you can imagine is very difficult to see on a 4k screen.
Hi Marco, you can't screens shot a mouse pointer. I'd love to try the latest build, but unfortunately we're not allowed to use anything but the official build. Is there any chance you guys could update the official download so that it's the latest stable daily?
May 9 2019
Hi Brecht, we're not allowed to use daily builds. Any ETA on when the main download link will be updated to a more recent version that may contain relevant bug fixes?
May 8 2019
file attached. Use any hdri to light the scene, and use this texture from poligon:
May 7 2019
Apr 19 2019
Thanks Brecht, that's great news, as this is crippling workflow at the moment. Perhaps an option to only calculate selected objects during anim player playback would also be beneficial (making live playback as fast as manual scrubbing( dependant on what's selected))......and an instance array modifier for arrays that don't need to be deformed or result in one merged object....requested here:
one thing that might be a bug though (can't imagine it's by design)....when things are hidden, either by having the layer switched off, or the object not visible, is there any point in it's modifiers being calculated during anim player playback?
Scrap that, after searching the scene item by item (I didn't create the file), I found that there was one object nestled amongst several hundred that had an array modifier on it! Resolved the playback speed by replacing the array with dupli groups.....I pray for the day that we can have an instance array modifier like we have in Houdini.
Apr 18 2019
Apr 4 2019
file attached. Try moving the perspective whilst viewport rendering, the banding stays relative to the screen and not the perspective movement.
Mar 8 2019
Here's the problem, the adaptive subdivision is being carried out on the first instance only, so that the end of the first instance is low poly, but then it meets back up with the high poly end when it's repeated. Would it be possible to perform the adaptive subdivision as it currently does, but limit the adaptivity to the inner part of the mesh, so that the outmost loop of polygons all get the same number of subdivisions as whatever the highest value on the mesh is (closest to camera).
A fix would probably be to make sure a change in subdivision cant occur where two instances meet.
with seamless texture:
Hi, the object, mapping, and noise are identical on both left and right models. This is just an example file, the client file was using a seamless texture from poligon, and the results where the same. No matter how high the subdiv's are increased or how close the camera is, the result is still a visible seam between instances.
normal map works correctly:
texture is seamless between duplicates (see right hand object which has same material minus the displacement).
Feb 17 2019
Jan 31 2019
cool, thanks for letting me know. Looking forward to 2.8 next year!
Jan 29 2019
Jan 28 2019
Jan 23 2019
Ah sorry, it was such bad logic I presumed it must be a bug. Posted in the feedback forum. Apologies.
Jan 19 2019
perhaps it would be possible to have cycles to pause rendering automatically when the gpu becomes unresponsive for system tdr - 1? Maybe throw up a message asking if you'd like to swap to cpu mode, with a remember this decision for future occurances (overideable in the user preferences system tab)
Jan 18 2019
Hi Brecht, you're quite right, I've just checked the system log and it was gpu timeout.