Perhaps it would be useful to explore what the render codec is used for. When Liquid has to render effects or transcode formats on the timeline (i.e., any slice that is not gray) it must process the video material by reading in the source, converting it and then writing the result back to disk to turn the slice 'green'. A green slice is Liquid's way of telling you that it has done the processing and for that particular slice, it is no longer reading the source material from disk, but is now reading the processed or 'rendered' material (an entirely different file on your disk).

So what format is the rendered material? Well - for DV material, the default is a DV format. For an HDV timeline the default is HDV (MPEG-2). The problem with both of these is that you get generation loss of video information during the rendering process (both resolution and color information). To minimize the loss of information you have the option of switching your rendering codec to an uncompressed format (using the droplist in the timeline properties), however; this uses a huge amount of disk space.

Thomas' tool provides us with high bitrate MPEG2 codec settings that allow for render processing that maintains more color information (4:2:2) and higher bitrates so that we can gain benefits closer to using uncompressed rendering at a fraction of the disk space.

Once a timeline is rendered, you can extract the material from Liquid by fusing the timeline (which is the act of converting the timeline to a single file that is the format specified by the render/fuse codec). One nice thing that Liquid does with fusing is that it's pretty smart about using the material already rendered to green slices so for example, if your timeline was 90% green slices and the rest gray, then only the gray slices would be transcoded to the fuse/render codec during a fuse operation (the green-slice material would simply be copied). Of course, if your gray slice material was already in the render/fuse format (as with HDV on an HDV timeline) then the fuse directly copies the source material without transcoding.

salexander:

Thomas' tool provides us with high bitrate MPEG2 codec settings that allow for render processing that maintains more color information (4:2:2) and higher bitrates so that we can gain benefits closer to using uncompressed rendering at a fraction of the disk space.

Steve, thanks for that explanation - really really helpful. One question - I thought HDV was only 4:0:0 - does this tool "upconvert", or is there something else you have to do to make good use of the information contained in a 4:2:2 stream?

Thomas

Thanks for the reply. This is what I thought you were trying to accomplish (and a very good thing it is). But I see no new choices when a try to create a new sequence. I've probably not installed to the correct directory, yes?

And no changes otherwise to my workflow. I still convert to MXF with the Sony software and then XReceive to my timeline.....

lasssman:

salexander:

Thomas' tool provides us with high bitrate MPEG2 codec settings that allow for render processing that maintains more color information (4:2:2) and higher bitrates so that we can gain benefits closer to using uncompressed rendering at a fraction of the disk space.

Steve, thanks for that explanation - really really helpful. One question - I thought HDV was only 4:0:0 - does this tool "upconvert", or is there something else you have to do to make good use of the information contained in a 4:2:2 stream?

Actually HDV is 4:2:0 I believe. The point is that if you render to an uncompressed color space (4:4:4) then all subsequent processing will continue on that color space (e.g., if you fused your sequence to a 4:4:4 color space and then brought it back in for further processing in a 4:4:4 color space). In theory, working this way you would be able to maintain excellent color integrity. Prior to Thomas' tool, you had either the choice of working in a 4:1:1 (DV) or 4:2:0 (HDV) color space or in RGB/YUV uncompressed. With his tool supporting 4:2:2 it's not quite as good as uncompressed but it takes up significantly less disk space.

Add to that the fact that an I-frame MPEG-2 with a sufficient bit rate will closely approximate the MJPEG codec in terms of quality, but you can't fuse to the MJPEG codec. I used to tweak my own fuse codec to 50mbps I-frame MPEG-2 and wished that I could push the bitrate higher to get better quality from my rendered effects without having to take the disk-space hit from going uncompressed.

Just to compare, try creating a 1080i60 project 1 minute long and fuse it to either RGB or YUV uncompressed and see how huge that 1 minute file is. It's huge!

Cheers

salexander:

... if you fused your sequence to a 4:4:4 color space and then brought it back in for further processing in a 4:4:4 color space ...

So do you set up a timeline with this new CODEC for subsequent processing? Or do you bring the fused clips back into the HDV timeline?

If I were to fuse out a 4:4:4 timeline I would likely reimport it into a 4:4:4 timeline in order to perform any additional processing. Think of the Mercalli workflow where you might remove the shake from some HDV video placed on the 4:4:4 timeline. Fuse it out and then import it back into Liquid to merge it with other video on another 4:4:4 timeline.

BTW - All this talk of 4:4:4 is somewhat misleading. In terms of the tool Thomas gave us, you might use a 100mb 4:2:2 codec to get the best compromise between disk usage and render quality, but the general approach is the same.

In my workflow, I likely will never go back to a timeline with the standard HDV render/fuse codec because when I finally come to export my sequence, I will either fuse it out and then use the result in some other tool downstream or I will export it with the MPEG-2 (or other) export mechanism.

It's also a little known fact that if you start with an HDV 1080/59.94i timeline and switch the render/fuse codec to uncompressed, for example, add a few effects here and there and then export to MPEG-2 with the HDV 1080 preset, Liquid is pretty smart about the export. It only re-encodes those bits of the stream that require rendering. Any native HDV 1080 material on the timeline will essentially just be copied out during the export. This is only true if you use the MPEG-2 HDV 1080 preset with a 1080/59.54i timeline (never tried it with PAL 50i but that likely has a similar preset). Disclaimer: I get exports at much better than real-time with the technique described here. It is my belief that Liquid could not be re-encoding the native HDV material at this rate and hence my assertion that Liquid does not re-encode the native HDV material when using this preset

I'm very grateful to Thomas for the liquid tool and to Guy for asking the question that less accomplished users of Liquid like me wanted to ask but didn't.

Please feel free to dumb the conversation down as you are doing. I think it's getting down to my level of understanding but I need some time to sit down and think it through. Perhaps this weekend.

Thomas,

 

I used the EX1 specific codec for this trailer.  I am really excited about the visible improvement in output.  I am now able to see the full potential of the EX1.

(trailer is 89mb)

Thanks for sharing this!

 

Mervin

Just watched it Mervin, real nice, a VBR 4,8 mbps mediafile but wasn´t some parts a little out of focus, I have heard that the EX1 might be a little tricky to setup at the first start, Thomas codecs seems to work well, in what way did you use them in your workflow.

Also, what app was used for the mediaencoding, did you do it directly out from Liquid.

thanks for the feed back.  I am so glad it is working out well for you.

I can't wait to show you guys my next tool which can up convert EX1 and HDV material to I frame only 4:2:2 m2v files.  Believe or not the video portion of it works great but I am still having problems with the audio side of it.

Hi Tomas,

It looks really sharp to me,  I don't use any sharpening in the presets but it looks very focused.  There is a shallow depth of field in some shots but that is by design.  I edit in Liquid and export to Sorenson.

 

Mervin

With all respect Mervin, I am only trying to help, I have to disagree.

I have seen great shots from that cam, this is a good one I think but this and this are less good.

As you mentioned going to bluray in the trailer I must tell you that all this will be visible, sorry but just for your knowledge, better you know it before the customers will discover it.

What codec from Smetvid did you use for your timeline?

Before the intent of this thread diffuses gently out to sea, I'm clearly missing something obvious: How do I use (and where should I be looking for) the new presets once I've installed them? What's the workflow? Example: I import material recorded on SxS cards onto my video drive, next I...........

Sorry I'm being thick.

Hello Tomas,

 

I don't mind your feedback at all.  But you are extracting those stills from a highly compressed file.  Take a look at some of the stills on my website (they are taken from the HD version directly from clipbrowser). 

 The camera is hard to set up and I sometimes think I should use a little bit of sharpening but in this case I am editing with an HD monitor and they look astoundingly sharp.

 

Mervin

Ok

Hello to All I have unloaded the program I wanted to know which it was the installation distance and which rows it is useful to make a backup, which rows must be sovrascrivere. E' possible to have a Tutorials little one for l' installation. In the event the translation dall' English all' Italian has confused something Thanks

Thanx for sharing your knowledge and codecs!

I've made a web site where people can see the content using broadband/ browser/ flash. 
This is how I do it;
- Record in SD format (have a HDV camera, but are recording in SD)
- Import to Avid 7.2 for editing (AVI format, codec DFI)
- Fuse to a uncompressed AVI file (makes huges files, but I delete them when I'm done)
- Code the exported file to Flash with On2 Flix Pro (VP6-E, 2 pass, CBR, 800 kbps, 1024 x 768)

So the question is, would the codecs you have made help me get better quality?
(or can anyone see something I should do different?)

jan-helge@exerto.no:

)

So the question is, would the codecs you have made help me get better quality?
(or can anyone see something I should do different?)

Jan Helge, those codecs are for rendering hd media and this thread is going rather Ot right now but I will make a little try.

As you have a HD cam , you should of course shoot and edit in HD , in that case, you can use those codecs as Thomas has described in the beginning of the thread to get better rendering of your HD media , your export settings out from your rendered Hd project could be a fuse , or , if your encoding app not is able to import those fused files , export as an 1080 x 1920 HD mjpeg AVI file , you will need to create your own preset for that in the export as AVI window .

What shel i say you did a great job

 

thank you very much keep on working

 

 

manu

salexander:

Disclaimer: I get exports at much better than real-time with the technique described here. It is my belief that Liquid could not be re-encoding the native HDV material at this rate and hence my assertion that Liquid does not re-encode the native HDV material when using this preset

 to verify your assumption, you can try this:

launch the Windows Task manager and watch the CPU Usage History graph (Performance tab) (suggestion: set the View/Update speed to low)

• when Liquid is NOT transcoding, the CPU(s) are almost idle (below 50%), since the fuse is I/O bound
• when Liquid is transcoding, the CPU(s) are at or near 100% (and you disks lights are flickering intermittently) since transcoding is CPU bound

 

mmmmmhh

 

Isn't this tool an answer to my headache here: http://forums.pinnaclesys.com/forums/thread/229416.aspx and here http://forums.pinnaclesys.com/forums/thread/253142.aspx ?

by using one of these new codecs (and converting my AVCHD clips to MPEG-2 using the same settings), I could fuse without transcoding ?

I will try soon, but a comment would be welcome anyway

Just remember to try if the fused file is able to use in the app that you want to import it into, Benoit , some settings doesn´t import well .

Thanks for the tool, but I am having a problem. Once the video is finished rendering it will not playback on the Timeline. If I put another vid on the timeline, which does not require rendering it plays fine but then I go back to any rendered section and they wont play. I am using Sony AVCHD 1920x108i 29.97fps files that I converted to MPEG2 50Mbit in Studio 12. (by the way, this works beautiful) I have the Chrome set to your 160Mbit render codec. But any of your codecs I use will not play. The 3 stock Liquid codecs play back fine. Any ideas?

Thnx

Roland

 

VM2 Productions:

Thanks for the tool, but I am having a problem. Once the video is finished rendering it will not playback on the Timeline. If I put another vid on the timeline, which does not require rendering it plays fine but then I go back to any rendered section and they wont play. I am using Sony AVCHD 1920x108i 29.97fps files that I converted to MPEG2 50Mbit in Studio 12. (by the way, this works beautiful) I have the Chrome set to your 160Mbit render codec. But any of your codecs I use will not play. The 3 stock Liquid codecs play back fine. Any ideas?

Thnx

Roland

 

what happens if you put in a still frame or any other type of video and render with one of my codec presets?  Does it still not play back?  I suspect it may have something to do with how the AVCHD files are converted in Studio 12.  I recently have been trying out some material from a beta version of the Flash XDR device and have had problems like that.  The files from the XDR were in a quicktime wrapper and I used an open source program that stripped out the raw m2v data but it was still a bit messed up even though Liquid saw it as a proper m2v file.

BenoitM:

Isn't this tool an answer to my headache here: http://forums.pinnaclesys.com/forums/thread/229416.aspx and here http://forums.pinnaclesys.com/forums/thread/253142.aspx ?

by using one of these new codecs (and converting my AVCHD clips to MPEG-2 using the same settings), I could fuse without transcoding ?

I will try soon, but a comment would be welcome anyway

Well...it doesn't solve my problem: feeding a 35 Mbps 1920 * 1080 / 50i CBR MPEG-2 file (converted from AVCHD by TMPGEnc Express 4) in the new timeline (Codec #(8) = 35 Mbps IPB CBR) yields gray slice but full re-encode at fuse

I could not spot any difference between the input clip and the fused output, except the actual bitrate: 33 Mbps for the input clip and 33.6 Mbps for the fused output (as reported by mediainfo; this tool also reports an identical nominal bitrate of 35 Mbps for each of the clips). In particular, GOP structure is almost identical (almost, because GSpot reveals an unexpected discontinuity in the sequence of GOPs for the fused output that does not exist in the original clip)