chocomel wrote:to test if the "judder" is related to your display config you can check out this page:
https://www.testufo.com/photo#photo=que … ;height=-1
it's very easy to see if your display can handle the framerate
Just close everything else like Browsertabs or Programs that are Ressource-Intensive and/or use your GPU in some way and then press the fullscreen Button in the upper right.
Also Edge and IE are not supported. FF, Chrome or Opera are fine.
If it is completely smooth, it's probably your rendering times that cause the judder. Meaning your GPU/CPU can't handle decoding/processing/rendering of your source file + SVP.
If you see interruptions, your display can't handle the framerate.
Also i saw you did the custom resolution via NVCP... while in theory that should work its probably a good idea to check if the settings madVR calculates for you give a better result.
Your 'Power management mode' in the NV Settings should be 'Adaptive', also try to force VSync: ON for testing purposes - or better create a Profile for your video player (to prevent issues with other programs or games) and leave the VSync setting on Application-controlled.
The Dell is set as primary display, with PJ 2nd, extended displays.
Monitor connected via DVI, PJ via HDMI.
The VGA has an 2.0b HDMI and the PJ has an 2.0a for HDMI port #1 and 1.4a for HDMI port #2 (regular 10m/30ft cable), so I don't think bandwidth is a problem.
Tested both PJ HDMI inputs.
Test UFO seems to require the test be run on the primary display, if the test is run on the 2nd display it will take the refresh rate from the primary.
Tested on FF x64, latest, all add ons disabled.
The Dell ran smoothly on 4 refresh rates: 48, 60, 72, 75.
The Benq ran smoothly on 2 refresh rates 48 and 60. On 72 and 75 there was constant judder.
I ran the video test file again with the refresh rates that were smooth on the respective displays, and judder was still present.
Tested SVP on the displays while they were primary/secondary (while extended). Also disconnected each one to see if it would make any difference with only 1 display.
In madvr OSD, the render time is about 8ms, so under the required vsync.
Also in madvr OSD the screen refresh rate is accurate, with one exception: with the Dell on fullscreen (it says it's 60Hz). To achieve a correct value Full Screen Exclusive needs to be disabled.
When testing SVP with 48Hz the CPU is at 50% and GPU at 40%, with judder present. On 60Hz CPU is at 70% and GPU at 90%.
Updated drivers from 416, to 418 and now to 425.31 (last driver with 3D support). Used DDU.
Tested with vsync on (was set at Use the 3D application setting) and changed PMM to Adaptive (was set at Prefer maximum performance). No difference.
The GPU can do PCIE 16 x3, tested with GPU Z.
Settings in NVCP for PJ are: RGB, Full dynamic range, 8bit color depth. PJ is set to PC signal.
Tested with RGB Limited as well (changed from NVCP and PJ). YCbCR is not available for 72 and 75 Hz.
PJ settings: Brilliant color Off. Tried Cinema (Rec. 709) and Game modes. Keystone Off. Noise Reduction 0. Film Mode On. No overscan. HDMI Audio Off (from Windows, NVCP, PJ).
It does not appear all the time, just in some scenes.
For testing purposes I use Love, Death and Robots episode 2 (Morpheus) 3min:42sec, when the camera pans from right to left.
Downloaded the file. Tested at 72Hz on both displays.
There seems to be some minor judder/artifacts at second 4 (around the main character), and the next second 5 on the left side of the face (around the ear).
The rest is too fast to notice anything wrong, but not very smooth.
dlr5668 wrote:also https://testufo.com/frameskipping + camera. SLR is the best but even cheap 80$ android is OK. Setup lowest ISO + continuous shot. Grab 50-100 shots and check them all
Will try as soon as I can get a camera.