BD-Rebuilder, Feature Requests

[stenews]

@jdobbs
thanks for answering me
and so the best thing to do is always to stop it at the end of one encoded segment. isn't it?

P.S.
same thing when I use DVD-RB...right?

[jdobbs]

Originally Posted by stenews

@jdobbs
thanks for answering me
and so the best thing to do is always to stop it at the end of one encoded segment. isn't it?

P.S.
same thing when I use DVD-RB...right?

Exactly -- except the segments in DVD-RB are typically much shorter. The best time to stop is right after the end of either of the two passes (assuming two pass is selected).

[stenews]

Thanks a lot jdobbs, I appreciate that.
You're doing a really great job with those applications!

Bye,
Stefano.

[Sn8kbordin]

Ability to select different path for work files and final output.
This way the rebuild process will be sped up if its on 2 different drives.

Thanks jdobbs for your great work.

[turbojet]

From my experiences CRF saves bitrate for the main movie while retaining a certain level of quality and it also saves some time but the current quick CRF uses settings aren't so good for keeping quality. Any chance of having separate CRF quality options?

Also setting quick CRF results in much larger under/over sizing, usually 500-1500 MB for BD5/9 outputs. 2 pass on everything is usually within 200 MB in the latest versions for me. Can the CRF sizing be looked at?

Also would like to resize the extras to 480 or 720 but keep the main movie at 1080p with an option to set M2TS to not resize by right click. Another thing could be an auto-resize option which would encode the extras and then encode 3% or so (using selectevery) of the main movie at CRF 26 (adjustable in ini). If the bitrate is equal to or less than the bitrate calculated from the remaining space it would keep it at 1080p, if the bitrate is more then it resizes to 720p.

[Dark Shikari]

Why not do this, jdobbs?

1) use --pass 1 --slow-firstpass on the CRF pass
2) If it oversizes, run a second pass.
3) If not, you're done...

[jdobbs]

Originally Posted by turbojet

From my experiences CRF saves bitrate for the main movie while retaining a certain level of quality and it also saves some time but the current quick CRF uses settings aren't so good for keeping quality. Any chance of having separate CRF quality options?

Also setting quick CRF results in much larger under/over sizing, usually 500-1500 MB for BD5/9 outputs. 2 pass on everything is usually within 200 MB in the latest versions for me. Can the CRF sizing be looked at?

Also would like to resize the extras to 480 or 720 but keep the main movie at 1080p with an option to set M2TS to not resize by right click. Another thing could be an auto-resize option which would encode the extras and then encode 3% or so (using selectevery) of the main movie at CRF 26 (adjustable in ini). If the bitrate is equal to or less than the bitrate calculated from the remaining space it would keep it at 1080p, if the bitrate is more then it resizes to 720p.

I'll add a hidden option to allow you to set the CRF value for the "quick" encodes. I'll look at the extra resizing and auto-size...

[jdobbs]

Originally Posted by Dark Shikari

Why not do this, jdobbs?

1) use --pass 1 --slow-firstpass on the CRF pass
2) If it oversizes, run a second pass.
3) If not, you're done...

Actually I'd planned to implement this on the one-pass CRF mode. It does a CRF size prediction followed by "--slow-firstpass" encode right now (so I can get the stats). The idea was to use the stats for a second pass in the event of an oversize -- but I hadn't gotten around to implementing yet.

[turbojet]

Originally Posted by jdobbs

I'll add a hidden option to allow you to set the CRF value for the "quick" encodes. I'll look at the extra resizing and auto-size...

Do you mean QUICK_CRF? That worked last time I tried it but --preset ultrafast takes away a lot of the advantages of CRF especially when aiming for BD5/9. I'd be content with CRF using the same x264 settings from the quality settings option. Maybe better would be a separate good/better/high/highest quality setting for CRF.

Thanks for looking at the resizing, I'd really make use of resizing the extras to 720/480, most don't have the definition you'd except from HD, to make for a higher quality movie.

[Sharc]

Originally Posted by turbojet

..... I'd be content with CRF using the same x264 settings from the quality settings option.

I second this.

Btw I did some tests with resizing the main movie anamorphic to 1440x1080; --sar 4:3 as a 'compromize resolution' for BD5/9. Worked pretty well, including subtitles.

[jdobbs]

Originally Posted by turbojet

Do you mean QUICK_CRF? That worked last time I tried it but --preset ultrafast takes away a lot of the advantages of CRF especially when aiming for BD5/9. I'd be content with CRF using the same x264 settings from the quality settings option. Maybe better would be a separate good/better/high/highest quality setting for CRF.

Thanks for looking at the resizing, I'd really make use of resizing the extras to 720/480, most don't have the definition you'd except from HD, to make for a higher quality movie.

If you use the One-Pass CRF it does use the settings of the Quality selections.

The problem is that if I use the same settings for QUICK_CRF, it suddenly isn't "QUICK" anymore. What do you lose in ultrafast that is directly related to CRF?

[setarip_old]

@jdobbs

Is there a specific reason that you've not implemented "drag & drop" of the source folder and the work folder?

[turbojet]

Originally Posted by jdobbs

If you use the One-Pass CRF it does use the settings of the Quality selections.

But I mean extras at CRF which gains me 1000-2000 kbps for the main movie usually using CRF 24 for the extras.

The problem is that if I use the same settings for QUICK_CRF, it suddenly isn't "QUICK" anymore. What do you lose in ultrafast that is directly related to CRF?

Better quality and lower file size. I don't mind the speed drop as it would still save a lot of time over 2 pass and it saves a nice amount of bitrate for the movie and I know what kind of quality it's going to be. 2 pass extras sometimes creates a saturated video stream to me (CRF <20) and occasionally results in some pretty bad looking video (CRF >28) because of how it's distributing the bitrate between the m2ts's. CRF solves this issue.

Another thing I thought of is extending the auto-resizing option with a suggestion for resize and file output size. For example:
- Set DVD5 output, no resize
- Encode all extras at CRF
- Sample ~3% of the movie at 1080p with CRF, compare to available bitrate, if it's below continue with the encode
- If CRF bitrate is above do CRF sample at 720p, if below pop up a box stating 'After analysis suggest to resize to 720p (default) or encode at 1080p'
- If CRF bitrate of 720p DVD5 is above, calculate bitrate for 1080p DVD9, sample and if below suggest between '1080p DVD9 (default), 720p DVD5, 1080p DVD5'
- If CRF bitrate of 1080p DVD9 is above, sample at 720p DVD9, if below suggest '720p DVD9 (default), 1080p DVD9, 720p DVD5, 1080p DVD9'
- if CRF bitrate of 720p DVD9 is above, suggest using 1080p BD25

Enabled by 2 options, auto resize and auto file output. Also have an option to decide which sets the default without user interaction. ini setting of AUTO_CRF or something with default at 26.

It would make the decision a lot easier and better for both novice and experienced users when aiming for DVD5/9.

[Sharc]

In my experience higher resolutions tolerate higher CRF values for same visual quality regarding encoding artefacts. How to account for this with auto-resize?

[turbojet]

True, what do you think of 720p CRF + 2 = 1080p CRF?
eg: 720p CRF 24 = 1080p CRF 26

Instead of crf samples a much better decision might be made from the rate factor of a full first pass. But that might have some disadvantages when resizing for second pass with the I/P/B changing. I've used 1080p stats files for 720p and vice versa a few times and it's always been within 10 kbps of the target bitrate with nothing noticeably wrong with the video. But maybe there's something technically bad by doing it.

Maybe the popups to resize are a bit much, and when auto resize and/or auto filesize is enabled it decides by itself but never exceeds DVD9 output.

[Sharc]

Well, personally I like to keep control over the resizing, rather than leaving it to some automatism. But that's just my opinion.
I would however be positive about the possibility to select the resizing for Extras and Feature independently (similar to DVD-RB).

[jdobbs]

Originally Posted by Sharc

Well, personally I like to keep control over the resizing, rather than leaving it to some automatism. But that's just my opinion.
I would however be positive about the possibility to select the resizing for Extras and Feature independently (similar to DVD-RB).

That'll happen, but its probably quite a way down the road.

I added a new hidden option for the next version:

QUICK_USE_QUALITY=1

This will use the selected quality option for quick encodes, but the only speed advantage you'll see is the elimination of the a pass (assuming two-pass mode). If you are using "One-Pass CRF" mode, the only advantage you'd see is elimination of the prediction passes.

Also, just a reminder to anyone reading this:

QUICK_CRF=n

will set the quality level of QUICK encodes to the CRF value you set in "n". So between the two you should have complete control over the quality of the "Quick Extras".

[Capsbackup]

IMHO, I am all for "tweaking", but I have found the default settings are hard to improve upon. I have to say I like the method of rate control of DVD-RB, simple slider/percentage style.
I understand the need/desire to satisfy the "technically advanced tweakers" out there, , but I like the simple approach as well. Hidden settings may work just fine for some, but I think the majority of users will benefit much more from simpler gui adjustments.

[Sharc]

Originally Posted by jdobbs

I added a new hidden option for the next version:

QUICK_USE_QUALITY=1

This will use the selected quality option for quick encodes, but the only speed advantage you'll see is the elimination of the a pass (assuming two-pass mode).

Nice option, thanks. The speed advantage will still be there in particular for larger Extras.

If you are using "One-Pass CRF" mode, the only advantage you'd see is elimination of the prediction passes.

Not sure if I understand 'elimination' correctly.. Don't we need a CRF prediction pass always for 'One Pass (CRF) encoding' and '"Quicker" encode for Extras' -- even for the Extras? Or are the Extras encoded at a fix (default) CRF?

Added:
I guess I found the answer: For 'One Pass encoding' AND 'Quicker Encode for Extras', the Extras are encoded at a fix CRF depending on Target Size or on the QUICK_CRF=n setting in the config file.

[jdobbs]

Exactly. In QUICK encoding a fixed CRF is used, and the output size of the non-extras are adjusted appropriately (larger or smaller -- normally larger, but it's up-in-the-air when you select something with QUICK_CRF) depending upon how much space is actually used by the combination of all the QUICK encodes.