Can anyone else confirm an issue with the Mac version where the "No DCT Decimation" is not saved in any preset?.
To reproduce, create any preset with the option ON. Reload the preset, and it will always be unchecked (OFF).
I've even tried editing the preset file itself and it won't accurately reflect the no-dct-decimate=1 in the option string.
V 1.0.7 (2017040900)
Mac OX X V 10.12.4
No DCT Decimate won't "stick" in settings?
Forum rules
An Activity Log is required for support requests. Please read How-to get an activity log? for details on how and why this should be provided.
An Activity Log is required for support requests. Please read How-to get an activity log? for details on how and why this should be provided.
Re: No DCT Decimate won't "stick" in settings?
Perhaps try adding dct-decimate=0 in the additional options box.
An activity log and screenshot would help.
An activity log and screenshot would help.
Re: No DCT Decimate won't "stick" in settings?
You can select it after loading the preset and it will add it to the option string without issue. I'm not sure what value a screenshot would provide as it would just show the box as unchecked. The only issue here is when saving preset with the option on (checked), and then using the preset, it always defaults to OFF again. I think tried editing the preset directly, and it still seems to ignore that as well. Once you load the preset though, you can select it and it will allow you to check it and encode with it on. A minor nit, but when encoding older noisy/dark video, it generates too much banding with it off, so every time I have to go in and re-check it each time I queue up another video.
I can't launch another instance of the encoder at the moment. I'm in the middle of a large batch of videos. I'll have to wait until I'm home to launch it on my other Mac for the activity log.
I can't launch another instance of the encoder at the moment. I'm in the middle of a large batch of videos. I'll have to wait until I'm home to launch it on my other Mac for the activity log.
Re: No DCT Decimate won't "stick" in settings?
Okay. We'll have to take a closer look then. Please note the advanced video tab is deprecated, so we definitely recommend using the advanced options box to enter the string manually going forward.
You may also want to consider aq-mode=3, which biases the adaptive quality algorithm to allocate slightly more data to dark areas of the scene. I've only found it to be slightly helpful on the most dark scenes, so your mileage may vary.
One last thing. Most of the time we hear about extreme banding, it's because the person's display isn't properly calibrated, increasing the brightness and contrast of the extreme black levels, which are meant to be very close to each other and not far apart in value. Please make sure your display is set up properly. There are some test charts here that may help verify your tone gradation is smooth: http://www.lagom.nl/lcd-test/all_tests.php
You may also want to consider aq-mode=3, which biases the adaptive quality algorithm to allocate slightly more data to dark areas of the scene. I've only found it to be slightly helpful on the most dark scenes, so your mileage may vary.
One last thing. Most of the time we hear about extreme banding, it's because the person's display isn't properly calibrated, increasing the brightness and contrast of the extreme black levels, which are meant to be very close to each other and not far apart in value. Please make sure your display is set up properly. There are some test charts here that may help verify your tone gradation is smooth: http://www.lagom.nl/lcd-test/all_tests.php
Re: No DCT Decimate won't "stick" in settings?
Somehow I missed this reply, so apologies for that. I'll give the AQ setting a try. I am curious why you are deprecating the Advanced video tab? It allows us to tweak various settings that are otherwise hidden via profiles. I would think software like Handbrake would understand the need for this? I'm guessing this functionality will be retained either via the custom strings, or some other option to tweak these going forward? This is probably not the proper forum for that discussion though.
On the topic of banding, the display is properly calibrated. These are less about banding (although that is a factor) and more about 'sparkles', where you get scattered macroblocks in dark areas that are of a slightly different luminance than the surrounding macroblocks, so it appears to 'sparkle'. Usually this takes the form of banding of smooth gradient areas, but on complex dark backgrounds, this sparkling can be pretty irritating with an almost mosquito noise like quality but made of larger macroblocks rather than individual pixels. I think one of the worse sources I have for this is an old DVD of Dragonslayer, which I've found to be nearly impossible to eliminate these to the point where I just ripped the source at a 1:1 copy and left it in an MKV container. It's typically more obvious on older source material from DVD's and pre-digital sources, but can still be seen even on newer content like Prometheus. The other odd thing is that the option string appears to contradict what's selected for DCT Decimate. Here's the resulting option string after I've set the options that I typically use for a BD rip.
trellis=2:subq=10:psy-rd=1,0.2:b-adapt=2:dct-decimate=0:ref=8:direct=auto:bframes=16:merange=32:me=umh:analyse=all:no-dct-decimate=1
Note the 'dct-decimate=0' and then the appended 'no-cdt-decimate=1' that's added when I check off "No DCT Decimation". I've been unable to find a setting for "dct-decimate", so I'm not sure if this directly contradicts the 'no-dct-decimate' option that I choose or not.
In short, I'm less concerned about getting the smallest possible size and retaining some fine grain/details in darker area. DCT-Decimate is of course the first place to look. I'll certainly give this AQ-Mode=3 setting a try as well. I wasn't aware of this one. Hopefully some good results from that!
Thanks for replying. It's really appreciated, as this is by far my go-to tool for almost every encoding job, and this is the one thing I've been unable to resolve in all these years.
On the topic of banding, the display is properly calibrated. These are less about banding (although that is a factor) and more about 'sparkles', where you get scattered macroblocks in dark areas that are of a slightly different luminance than the surrounding macroblocks, so it appears to 'sparkle'. Usually this takes the form of banding of smooth gradient areas, but on complex dark backgrounds, this sparkling can be pretty irritating with an almost mosquito noise like quality but made of larger macroblocks rather than individual pixels. I think one of the worse sources I have for this is an old DVD of Dragonslayer, which I've found to be nearly impossible to eliminate these to the point where I just ripped the source at a 1:1 copy and left it in an MKV container. It's typically more obvious on older source material from DVD's and pre-digital sources, but can still be seen even on newer content like Prometheus. The other odd thing is that the option string appears to contradict what's selected for DCT Decimate. Here's the resulting option string after I've set the options that I typically use for a BD rip.
trellis=2:subq=10:psy-rd=1,0.2:b-adapt=2:dct-decimate=0:ref=8:direct=auto:bframes=16:merange=32:me=umh:analyse=all:no-dct-decimate=1
Note the 'dct-decimate=0' and then the appended 'no-cdt-decimate=1' that's added when I check off "No DCT Decimation". I've been unable to find a setting for "dct-decimate", so I'm not sure if this directly contradicts the 'no-dct-decimate' option that I choose or not.
In short, I'm less concerned about getting the smallest possible size and retaining some fine grain/details in darker area. DCT-Decimate is of course the first place to look. I'll certainly give this AQ-Mode=3 setting a try as well. I wasn't aware of this one. Hopefully some good results from that!
Thanks for replying. It's really appreciated, as this is by far my go-to tool for almost every encoding job, and this is the one thing I've been unable to resolve in all these years.
Re: No DCT Decimate won't "stick" in settings?
The long and short of the advanced panel is, its a lot of work to maintain and most people, even enthusiasts, shouldn’t mess with those settings since x264 and it’s encoder presets are quite mature. The advanced options box is generally sufficient for minor adjustments.
The dark macroblock sparkles are what I was referring to; usually bad calibration in the near-black area. If your display is properly calibrated then perhaps employ the NLMeans denoiser.
The two dct strings are equivalent.
The dark macroblock sparkles are what I was referring to; usually bad calibration in the near-black area. If your display is properly calibrated then perhaps employ the NLMeans denoiser.
The two dct strings are equivalent.
Re: No DCT Decimate won't "stick" in settings?
Thanks and understood. In a situation like this where it has placed a string value and I later checked off the No DCT Decimate, do you know which would take precedence if any? I'll be sure to manually edit the strong going forward it it contains only the one entry.
Re: No DCT Decimate won't "stick" in settings?
Not entirely certain. The unparsed options string in the log may show it reduced to one value, or not.
Re: No DCT Decimate won't "stick" in settings?
Just an fyi, the aq-mode=3 did help with the dark areas. That combined with a CRF of 17 instead of 20 produced a very acceptable encode. I don't expect to need a CRF that low on newer sources, but I can mark this old analog copy off my list of 'difficult' encodes.
Thanks for you help man.
Thanks for you help man.
