View Profile
View Forum Posts
Private Message
I have what I think is a simple question. When I need to create a DVD, in the past I just exported out of PP using AME. However, this site has pointed me towards better encoders like HCenc. When I try to frameserve out of PP using DMFS it is painfully slow. However, if I export out of PP as a lossless AVI then encode that, it is probably like 2x faster. I know the whole point of frameserving is to avoid generating intermediate files. But I can push the delete button on my keyboard a lot faster.
So my question is specifically, do I introduce additional generational loss by exporting a lossless AVI then frameserve that to HCenc using Avisynth versus frameserving out of PP using DMFS to Avisynth then HCenc?
+ Reply to Thread
Results 1 to 18 of 18
-
-
Got my retirement plans all set. Looks like I only have to work another 5 years after I die........
-
Assuming the bitrate is normal it will not really matter which one you use.Originally Posted by SameSelf
What quality is the source? -
No additional loss for either method, if you keep the same colorspace / color model / color sampling
There are no compression lossess, but lossless codecs and frameserving are only "lossless" in the same colorspace. You can incur a colorspace conversion loss -
---Mpeg2 encoder is not that bad in Premiere or Vegas, maybe HcEncoder is better with low bitrate, not sure, I do not encode that way. Frame server helps out if original is HD interlace, because HD interlace to SD interlace sucks in Premiere or Vegas so using Avisynth gives you much better result and just HcEncoder loads Avisynth
---other reason for frame server, to save some bitrate using CQ (Constant Quality) in HcEnc , which is 1pass, if DVD is shorter than 1 hour and 10 minutes or so , CQ with max 8000 or so audio 256kbps, is fine, if you use Premiere you have to go with CBR (which is no problem for video not that long, if mpeg2 is not dedicated to sit on hardisk but DVD) or use 2pass, BUT 2pass takes longer,..., just saying that because you talk about encoding time .... so it is not like one have to blindly use this or that, ..., just to weight a situation and act upon it ...Last edited by _Al_; 25th Feb 2015 at 18:31.
-
Good stuff as always guys (and gals just in case I offend someone). Colorspace is something I am only just now getting my head around. As for the source it is mostly HD but I need to prepare a DVD for displaying on a projector for a presentation. Sadly few projectors are HD. But I want it to look as good as possible. The DVD is only like 20 minutes so bit rate is not a concern.
-
As for frame serving HD resolution through frame server and getting SD resolution, using YUY2 debugmode frame server export in Avisynth:Originally Posted by SameSelf
but careful , in Sony Vegas ( not your case) one has to fix levels to not have colors washed out, stretching it back to 0-255, which is not ideal, you might start to loose tiny bits of details in highlights, overburned areas exporting RGB through framesever, Vegas uses that color space internally and that is a base from frameserving, not a reason to change it, better in Avisynth , but it is not 100% guaranteed because Vegas can treat formats differentlyCode:AviSource("frameserver.avi") ConverttoYV12() #perhaps deinterlace if needed ColorMatrix(mode="Rec.709->Rec.601", clamp=0) #whatever filters
but one can just leave ConverttoYV12() if washed out colors are not a concern ...Code:AviSource("frameserver.avi") ConverttoYV12(matrix="PC.709") #perhaps deinterlace if needed ColorMatrix(mode="Rec.709->Rec.601", clamp=0) #whatever filters -
Good stuff _AI_. Has got me doing more research and looking back through all my footage. So please bear with me.
Starting with some Sonycam Hi8mm SD footage that I captured years ago via a Canopus ADVC-300 using S-Video in DV-AVI, Mediainfo tells me this:
Nothing in there about Rec.601. So I just assume that it is REc.601 YUV 4:1:1 720x480i29.97 bff 2.357 bpp with a dvsd FourCC. But is the colorspace [0,255] aka "PC" or [16,235] aka "TV"?Code:Video ID : 0 Format : DV Codec ID : dvsd Codec ID/Hint : Sony Duration : 45mn 22s Bit rate mode : Constant Bit rate : 24.4 Mbps Encoded bit rate : 28.8 Mbps Width : 720 pixels Height : 480 pixels Display aspect ratio : 4:3 Frame rate mode : Constant Frame rate : 29.970 fps Standard : NTSC Color space : YUV Chroma subsampling : 4:1:1 Bit depth : 8 bits Scan type : Interlaced Scan order : Bottom Field First Compression mode : Lossy Bits/(Pixel*Frame) : 2.357 Time code of first frame : 00:00:46;05 Time code source : Subcode time code Stream size : 9.12 GiB (95%)
If I write this out as a lossless AVI from PP I get the following:
So it looks like PP upped the Chroma settings: YUV 4:2:2 at 16 bpp. The Codec Info line is very interesting as well. Mediainfo doesn't seem to detect whether it is interlaced or not. Again, can't tell if it is Rec.601. It has a UYVY FourCC. But again, [0,255] or [16,235]?Code:Video ID : 0 Format : YUV Codec ID : UYVY Codec ID/Info : Uncompressed 16bpp. YUV 4:2:2 (Y sample at every pixel, U and V sampled at every second pixel horizontally on each line). A macropixel contains 2 pixels in 1 u_int32. Duration : 1h 1mn Bit rate : 166 Mbps Width : 720 pixels Height : 480 pixels Display aspect ratio : 3:2 Frame rate : 29.970 fps Standard : NTSC Color space : YUV Chroma subsampling : 4:2:2 Compression mode : Lossless Bits/(Pixel*Frame) : 16.000 Stream size : 70.8 GiB (99%)
I also have MPEG-4 AVC SD video (no capture involved) in a Quicktime container. This kicks off the following Mediainfo:
Well, ain't that strange? Seems to be mixing Rec.709 and Rec.601? No idea what the Colorspace is now.Code:Video ID : 1 Format : AVC Format/Info : Advanced Video Codec Format profile : Baseline@L3.1 Format settings, CABAC : No Format settings, ReFrames : 1 frame Format settings, GOP : M=1, N=15 Codec ID : avc1 Codec ID/Info : Advanced Video Coding Duration : 2mn 2s Bit rate : 9 901 Kbps Width : 640 pixels Height : 480 pixels Display aspect ratio : 4:3 Frame rate mode : Constant Frame rate : 30.000 fps Color space : YUV Chroma subsampling : 4:2:0 Bit depth : 8 bits Scan type : Progressive Bits/(Pixel*Frame) : 1.074 Stream size : 145 MiB (93%) Language : English Encoded date : UTC 2011-01-18 20:38:39 Tagged date : UTC 2011-01-18 20:38:39 Color primaries : BT.709 Transfer characteristics : BT.709 Matrix coefficients : BT.601
If I pull up my HD video, Mediainfo says the Color Primaries etc. is BT.709. So no mixing up HD.
But in none of these can I tell if I am looking at [0,255] or [16,235]. How do I know?
I found the following guides which were very helpful:
http://www.bubblevision.com/underwater-video/Vegas-YouTube-Vimeo.htm
http://www.bubblevision.com/underwater-video/YouTube-Vimeo-levels-fix.htm
http://www.glennchan.info/articles/vegas/v8color/vegas-9-levels.htm
So if I understand all this correctly, Avisynth filters only work in YV12. I am guessing this can be either Rec.601 or Rec.709 and either [0,255] or [16,235]. Therefore, it is crucial you know what you are exporting, try to minimize conversions. Can anyone provide a little more insight?
Thanks! -
Perhaps in Texas......?Originally Posted by SameSelf
Also if you are going to display this on a projector why on Earth convert to DVD and degrade the video, just plugin a laptop.
Also many of those old office projectors play S-video or composite really bad.
Folks how can anyone seriously suggest when a HD video needs to be played on a projector to go lossless intermediate to degrade it to SD then to degrade it again by re-encoding it to MPEG-2 , then put it on a DVD and hook up a DVD player to play it on a projector that is probably has a bad video circuit.
Just hook it up to a laptop, set the native resolution and play the damn video.
I mean am I missing something?
Last edited by newpball; 28th Feb 2015 at 13:01.
-
What type of project and types of manipulations and editing are you doing? Premiere can smart render DV. So if you are just doing cuts, then you should be exporting DV
NTSC DV is YUV 4:1:1 , but if you're going to DVD, eventually you will need 4:2:0
SD is always considered to use Rec601 by convention . Unless you know for certain otherwise, that's what you use. The metadata (or anything else) reported by mediainfo might be incorrect
Mediainfo cannot tell if the actual video data has YUV 16-235 or YUV 0-255 - it doesn't scan the file, only reads the header and metadata tags. It does however, report of a full range flag is preset. Again , this might or might not be correct. Just because something is flagged full range, doesn't mean the actual data is full range. Don't rely on mediainfo (it can be wrong) for these types of things
Mediainfo cannot tell if the CONTENT is progressive or interlaced. It can only tell how the video was encoded. For example, You can have progressive content encoded interlaced, or interlaced content encoded progressive. Don't rely on mediainfo for determining content
Basically, mediainfo cannot tell you anything for certain, because you can think of it as reading "labels" and metadata only, not actually scanning the file. But still it can be useful for some things
Yes, some avisynth filters work in YV12, but others can work in other colorspaces and there are workarounds as well to use other colorspaces and sampling
709 vs 601 only refers to the matrix used for the YUV<=>RGB conversion (e.g. when video is converted to RGB for display, or maybe when you are using RGB filters) . So if you used a full YUV workflow, this is irrelevant. It only matters , when you do YUV<=>RGB conversions -
newpball: Don't know about NorCal, but yeah, in the Texas public schools, SD projectors are all they got. I am just not a fan of bringing my laptop to these venues cause it is just unreliable versus popping a DVD into the player: don't want my laptop to break, suddenly malfunction, get stolen, etc. I don't even care if I get the DVD back. I can just walk up to the MC and say, "Hera ya go", and my job is complete. A lot of virtues are required in this role, can't demand anything really
pdr: My projects involve the following: color correction, unsharpening masks, titling, rolling credits, fade ins/outs, 2x speed ups, 50% slow downs, zooms, multiple layers of video, etc. Plus I am slowly learning how to do custom transitions in AE and adding those to my videos. The more I do this, the more I tend to push the limits because with a large audience as the target I like to spice up the video and engage the viewer.
Mediainfo unreliable. Gotcha. Well, I guess I can pull the video into AE and investigate RGB values there. I am not sure how to tell if video is interlaced or progressive. Is there a tool that does? This is a concern because a lot of video ends up on youtube as well.
709 vs 601 refers to YUV<=>RGB conversion. Got it. Maintain a YUV workflow and I am all good. Gotta look closely and make sure no RGB is slipping into my workflow. -
Yeah sure you must be the only person at those public schools who got a laptop.Originally Posted by SameSelf
And USB sticks, well, I suppose they can self explode, too much risk there as well.
No a DVD is gold, and the S-video and composite video processing circuity is marvelous in SD projectors, just make sure all your intermediate processing goes lossless, after all you aim for the best quality that's for sure!
Taking a HD video, degrading it to SD, re-encoding it to MPGE-2, putting it on a DVD and then playing it trough bad video circuitry on a projector and then worrying about YUV to RGB conversions?
Sound like a case of rearranging the deck chairs on the titanic!
Last edited by newpball; 28th Feb 2015 at 13:32.
-
pdr might respond if your mentioned premiere filters use RGB space,
DebugMode frame server has RGB 16, RGB32 and YUY2 which is 4:2:2 out, so exporting from Premiere to keep YUV , one has to go with that YUY2, but in Avisynth you have to go back to YV12 (4:2:0) using ConverttoYV12() , because that is what you need for general mpeg2 or H.264. What "damage" that causes I cannot tell exactly, I do not use Premiere.
Filters in Avisynth not necessarily need only YV12, some filters would take for example YUY2 as well, or using special dll's that are fixed for this. Some filtes are fine with RGB. Using virtual dub filters in Avisynth for example. So going from Vegas as to get intermediate, you can tweak that lossless still in RGB and load it back to Vegas etc. YV12 is for final delivery, all players expect that basically, just for compatibility I'd export everything 4:2:0. Storage, archive is another story anything is allowed.
To get SD from HD source,
you just load uncompressed from frameserver into Avisynth (getting YV12) that is uncompressed, then uncompressed color space to SD change: ColorMatrix(mode="Rec.709->Rec.601", clamp=0) , then it is encoded from that still uncompressed to mpeg2, you think it is too much? I would not worry. I'd not overthink it, it is all uncompressed all the way to encoding. DVD is just final delivery, not storage. I'm getting Raspberry Pi 2 B, , next week, much faster versions then first generations, reviews are very positive. I will test how it stands DVavi and Kodi for example, these things are cheaper than cheapest media players. And there are packages , you simply load it with whatever you want, even some Win10 version soon perhaps. The time when we can play originals perhaps is coming (like purchasing something with Kodi preinstalled etc) . Those HTPC boxes will get simplified for large population I think, and they get cheaper. yOu know install app, soft whatever you want, not just dumb firmware for good. Whatever it goes, but I guess we might dust off those originals as I said. -
AE works in RGB. So if you import some animations from AE, lower thirds, graphical elements, etc...they will be RGB unless you converted them to YUV
Premiere has a YUV capable timeline, so if you apply RGB effects, or overlay some RGB elements, lower thirds etc.. only those sections will undergo RGB conversion . If you apply YUV filters, those sections will still be YUV
You can determine if the content is interlaced or progressive by using your eyes. Each field during some movement will show a different moment in time. So 59.94 fields /sec will show 59.94 unique moments in time. If you're looking at frames, you will see combing or hoizontal lines during every frame with motion. I usually use avisynth for this and look at separate fields but it doesn't matter what you use as long as it's a NON deinterlacing preview method. (e.g. if you use some media players, some will automatically bob deinterlace DV , AVCHD or other video formats, others will half rate deinterlace) -
I think I am slowly figuring this out. Using color picker in PP I see the following. Never understood before but I think I do now. PP rescales RGB 0,0,0 and 255,255,255 to luma values of 16 and 235.
Thanks for the heads up about AE using RGB. Didn't realize that either. -
YUV 16-235 => RGB 0-255 is basically what a REC or "TV" or "limited range" conversion is in terms of 8bit values. There are slight differences between 709 and 601 in terms of color
In contrast, YUV 0-255 => RGB 0-255 is a "PC" or "Full range" conversion. This is essentially what Sony Vegas uses in 8bit mode with Studio RGB. (that's Studio RGB, not sRGB)
RGB and YUV are non overlapping color models. 8bit YUV can contain more information than 8bit RGB. What you see in the preview is RGB (ie. it's converted to RGB for display, even if the timeline is working in YUV), so you are only seeing part of the picture in the preview
The thing to be careful about with DV (or any camera footage) is usable "superbrights" . Almost all DV camers record usable data in the range Y=236-255 , or the "illegal" range. So when working in after effects, that data is clipped, unless you work at 32bit bpc. Adobe products in 32bpc allow you to access YUV data before the RGB conversion so nothing is clipped -
Just to firm up my understanding of these things: 32 bpc = 8 bits for R G and B plus alpha, correct? So still in 8-bit mode.
-
Originally Posted by SameSelf
32 bpc (Bits Per Channel) = 32 R, 32 G, 32 B + 32 A
Quote
