a few questions about h264

[psyde]

hello. i've been experimenting with masternobody's build for the past week, and i'm happy with the results they give from a quality standpoint (file size isn't an issue in my case). i've used tmpgenc for personal encoding for over ten years now, even for hd, and am glad i finally decided to break of of my set ways. i have a few questions about the codec that i'd appreciate anyone could answer.

the presets - i know that they each have their own given settings when it comes to quality and efficiency, but do the speed presets also correlate to motion searching? in other words, does it act in the same way that tmpgenc's motion search precision does in addition to the settings they use?

when i encode a video preserving the source's color space (keep input colorspace setting), i can't use vlc's video effects on the resulting video (rotate, color adjustments, etc.), but when i use the "convert to yuv 420" setting, i CAN then use vlc's video effects. this makes no sense to me, as my SOURCE video is yuv 420 to begin with. is there a certain setting that the 'convert to yuv' uses that makes this possible? the reaosn i DON'T want to use the convert to yuv feature is that it changes the colors - they become a bit more saturated/contrasted. to be clear, i want to keep my input's colorspace, but also would like to be able to apply vlc effects to the video.

about two pass encoding - i'm using sony vegas. for two pass encoding, am i doing this correctly? i set it to 1st pass, and render the video as say "1st pass.avi". when it's done, i go back to the codec setup window, and select 'nth pass', and proceed to encode "2ndpass.avi for the final product. everything tells me i'm doing it right, but i just want to be sure.

another question about two pass encoding. is it possible to give the encoder settings that would target an average bit rate of say 50000, with peaks of 100000 when needed and a minimum bit rate padding to not go below say, 25000? i realize that there is a maxrate command, and i'm guessing that is what i would use for max, but what about a minrate command - i haven't read anything about one. does vbvbuffer size have anything to do with this? from what i've read (i don't remember much of it now lol), it has to do more with streaming purposes and/or file size. neither of those are issues for me, i'm using h264 for quality archiving. i'd like to do what i did with tmpgenc - set an average with peaks for min and max where it's needed.

i'm not too ken on using/trying the tuning presets, as i use a very high bit rate to begin with (55,000 two pass for 1080p 6fps), but are there any other command that i could use to ensure that i'm getting the highest quality output possible besides "--no-deblock"?

thanks for your time.

[_Al_]

When talking about conserving color space that is done by what and how, in Vegas, tmpgenc? Not sure what you do. Vegas does not have any presets, like to encode faster or slower.

When exporting in Vegas, you can select two pass right away, it will run both passes

you cannot set minimum bitrate

you do not have much tweaking available in Vegas while exporting, no CRF encoding (1pass, saving time and encoding to quality), just 2pass and therefore guessing average bitrate (where you end up overshooting probably most of the time, if talking about encoding to average 50000, 420 h264, HD video)

[psyde]

thanks for responding, but i'm actually talking specifically about masternobody's h264 codec/gui.

[_Al_]

If file size is not an issue for you, use CRF encoding. You limit bitrate with proper quantizer, not giving encoder a fixed number. Quality cannot be defined with define bitrate for every video.

Use --tune noise or --tune film , if you intend to edit your video later then you might use --fastdecode or --keyint 60 or less. Filesize will be getting bigger if using CRF.
This is the thing, you test 2pass, then you change one setting, like above mentioned and you can start testing again. Then with different video, you can test yet again for correct bitrate. Forget about 2pass encoding.

if colors get changed , it might be limited vs. full range problem

[psyde]

If file size is not an issue for you, use CRF encoding. You limit bitrate with proper quantizer, not giving encoder a fixed number. Quality cannot be defined with define bitrate for every video.

Use --tune noise or --tune film , if you intend to edit your video later then you might use --fastdecode or --keyint 60 or less. Filesize will be getting bigger if using CRF.
This is the thing, you test 2pass, then you change one setting, like above mentioned and you can start testing again. Then with different video, you can test yet again for correct bitrate. Forget about 2pass encoding.

if colors get changed , it might be limited vs. full range problem

appreciate your input. i'll give crf a shot.

[poisondeathray]

To answer some of the other questions -

the presets - i know that they each have their own given settings when it comes to quality and efficiency, but do the speed presets also correlate to motion searching? in other words, does it act in the same way that tmpgenc's motion search precision does in addition to the settings they use?

Yes, each preset has motion estimation , subpixel motion estimation . Some have overlap; for example "ultrafast" and "superfast" both use --me dia , or "very slow" and "placebo" both use --subme 10 . The motion estimation range is also increased in the later presets .

when i encode a video preserving the source's color space (keep input colorspace setting), i can't use vlc's video effects on the resulting video (rotate, color adjustments, etc.), but when i use the "convert to yuv 420" setting, i CAN then use vlc's video effects. this makes no sense to me, as my SOURCE video is yuv 420 to begin with. is there a certain setting that the 'convert to yuv' uses that makes this possible? the reaosn i DON'T want to use the convert to yuv feature is that it changes the colors - they become a bit more saturated/contrasted. to be clear, i want to keep my input's colorspace, but also would like to be able to apply vlc effects to the video.

The input colorspace is actually RGB, because it means "direct input into the plugin" . ie. Whatever the plugin is "seeing" immediately . Since vegas works in RGB, that's what you're getting. That's usually where are the color changes occur as well. Because vegas uses studio RGB, and slightly different conversion for RGB<=>YUV . Also there are several different ways in which you can convert RGB<=>YUV . Most common problems are differences between full/limited range, and Rec601 vs. 709

about two pass encoding - i'm using sony vegas. for two pass encoding, am i doing this correctly? i set it to 1st pass, and render the video as say "1st pass.avi". when it's done, i go back to the codec setup window, and select 'nth pass', and proceed to encode "2ndpass.avi for the final product. everything tells me i'm doing it right, but i just want to be sure.

i'm not too ken on using/trying the tuning presets, as i use a very high bit rate to begin with (55,000 two pass for 1080p 6fps), but are there any other command that i could use to ensure that i'm getting the highest quality output possible besides "--no-deblock"?

Disabling deblock actually reduces quality in most cases , until you get to much higher bitrates on most sources (much higher than what you are using) . A better way for most cases is to adjust the alpha and beta deblock values. BUT if you really wanted highest quality - regardless of filesize, as you say it isn't an issue - you should encode lossless

i realize that there is a maxrate command, and i'm guessing that is what i would use for max, but what about a minrate command - i haven't read anything about one. does vbvbuffer size have anything to do with this? from what i've read (i don't remember much of it now lol), it has to do more with streaming purposes and/or file size. neither of those are issues for me, i'm using h264 for quality archiving. i'd like to do what i did with tmpgenc - set an average with peaks for min and max where it's needed.

It doesn't work like you probably think. Maxrate is actually the highest rate that can enter the buffer; it's not really the "maximum bitrate" the way most people think it is.

Another way to set a minimum level quality, is to set a maximum quantizer --qpmax . Not recommended unless you know what you're doing . For most people and general use cases, just changing the CRF value is a better way to go. Lower values yield higher quality, larger filesizes. By the time you get to zero, it's lossless in terms of input/output . (So in your case whatever vegas is sending the plugin, assuming there are no other conversions)

[psyde]

very interesting and informative, thanks. i'll have to google up a bit more of some of the terms you threw at me, but i'm glad i have new avenues to pursue. and i would go lossless, but that's a little much - file size does matter at a certain point - i'll have to experiment some more with cfr instead of 2pass vbr. thanks again mate.

[pandy]

It is not clear what is your goal... some effects may require RGB color space and as such you will be unable to avoid multiple YCbCr->RGB->YCbCr conversions - if your goal is to preserve original quality then don't use filters or effects, if not then focus on restoring previous contrast (i assume you are see on your PC display limited and full quantization range which may lead you to wrong conclusion about contrast and saturation).
Multipass encoding is used only when your target must fit within some limitations (such as fit your video within 4GB space related to FAT32 filesize limit or you for example encoding for DVD - only then 2 pass has sense) - normally use CRF as already advised - don't sent arbitrary limitations on codec (such as 55Mbps) unless you must (for example your decoder support decoding up to some limited profile \ level \ other arbitrary criteria) focus on quality.
For most users default x264 tunes and presets are usually more than OK - there is plenty "guides" on web where people introducing "magic" settings supported by some doubtful "knowledge". More complex coding is not synonym for higher quality - remember that lossy codec focus on delivering same (not higher) quality with lower bitrate so highest quality gain is from raising bitrate or rather less restricting encoder - focus on quality not on bitrate or filesize (unless bitrate or file size is your goal).