AF-100 Wishlist/Suggestion Discussion Thread

[Starshine Video]

Standard AB / V mount battery option. I see this as a highly improbable due to the majority of people wanting to use existing Pana 170/200/150 batteries, but for those that already use these...

We aren't wanting to have the Anton Bauer or V mount be the Only Option. Just have an officially supported adapter plate ala the JVC HD110. People could still use the standard panny batteries if they wanted, and those of us with nice big ABs could use those. Its not an either or thing.

 

[Jan_Crittenden]

a large, high rez LCD with something akin to the 920k pixel LCD's of my D300/D300s

Okay, feeling stupid here, which D300 from whom?

 

[Jan_Crittenden]

In defense of Panasonic and the codec, I think you need to take the fight to Apple as other NLEs can edit AVC natively and in real time.

Excuse me, why would you want to edit any long GOP codec natively? Am I missing something or does everyone just want to degrade their image at every edit point?

Thanks,

Jan

 

[Ben_B]

Excuse me, why would you want to edit any long GOP codec natively? Am I missing something or does everyone just want to degrade their image at every edit point?

Thanks,

Jan

I haven't the slightest idea...it seems utterly absurd to me as well.

Okay, feeling stupid here, which D300 from whom?

I believe it's a Nikon.

I would like to see a high resolution viewfinder at least as good as the DMC-GH1. The Lumix people understood the necessity of a high-resolution viewfinder to allow photographers used to an optical viewfinder to make the transition to a mirrorless system, but it's also great for video. Even the GH1's LCD looks better than that on the HVX-200, etc, but the viewfinder is absolutely exceptional...even cameras with excellent LCDs like the Sony PMW-EX1 have horrible viewfinders.

 

[Erik Olson]

This from page one, just thought Id replay it.

And probably to the irk of many

Standard AB / V mount battery option. I see this as a highly improbable due to the majority of people wanting to use existing Pana 170/200/150 batteries, but for those that already use these...

I like the option of the I/O module that takes p2 and has AVC Intra. Heres my issue, I dont have spare endless hours to transcode a long gop codec to an I frame codec. Yeah its easy to do, and yeah its workable, I just dont have the time and if I did I know a million other ways Id prefer to spend it.

I didn't think about the modular recording media dock. That's P2 or SD. Obviously, we've beaten the fact that this should be an AVC-I100 camera horse to death, but if I could buy a $1500 P2 module that simply docks to the camera as an accessory, then I'm a really happy producer.

AB mount +1!

 

[Matt Gottshalk]

If the swing out LCD is anywhere near the excellent quality of the HPX300 then I would be happy.

 

[robertrogoz]

Excuse me, why would you want to edit any long GOP codec natively? Am I missing something or does everyone just want to degrade their image at every edit point?

Thanks,

Jan

Are you talking about mpeg2? Because if you are, your information is completely incorrect in respect of long GOP mpeg2 and FCP quality loss. This topic was discussed in depth on dvinfo.net several times. Steve Mullen wrote about it as well, you can google his article. If you edit mpeg2 long GOP (HDV, XDCAM EX, DVCPRO HD and such) native and render in ProRes you'll suffer less quality loss then transcoding into ProRes and re-transcoding it again during the export. Also this system allows you to keep the HDD space to manageable levels, which is a big issue with larger projects.

 

[Ben_B]

Are you talking about mpeg2? Because if you are, your information is completely incorrect in respect of long GOP mpeg2 and FCP quality loss. This topic was discussed in depth on dvinfo.net several times. Steve Mullen wrote about it as well, you can google his article. If you edit mpeg2 long GOP (HDV, XDCAM EX, DVCPRO HD and such) native and render in ProRes you'll suffer less quality loss then transcoding into ProRes and re-transcoding it again during the export. Also this system allows you to keep the HDD space to manageable levels, which is a big issue with larger projects.

H.264/MPEG-4/AVC is what she's talking about.

 

[Otis Grapsas]

Are you talking about mpeg2? Because if you are, your information is completely incorrect in respect of long GOP mpeg2 and FCP quality loss. This topic was discussed in depth on dvinfo.net several times. Steve Mullen wrote about it as well, you can google his article. If you edit mpeg2 long GOP (HDV, XDCAM EX, DVCPRO HD and such) native and render in ProRes you'll suffer less quality loss then transcoding into ProRes and re-transcoding it again during the export. Also this system allows you to keep the HDD space to manageable levels, which is a big issue with larger projects.

The marketing of the all those "workflow" codecs is so strong that sometimes people forget that for a single edit cycle, the original codec is always better. Certain workflow codec manufacturer even support their sales with manipulated tests.

Junk->timeline processing->output codec is always better than
Junk->uncompressed->timeline processing->output codec and a lot better than
Junk->any workflow codec->timeline processing->output codec

There could be other advantages though, the workflow codec could have lower cpu use. Putting that aside, the extra transcoding always reduces the output quality of a single stage workflow.

Jan is probably talking about multistage professional workflows though, where processed material (not EDLs) is transfered from edit station to edit station. That requires multiple decoding/encoding cycles so you don't want to do that on a long GOP codec. Many codecs do not even support encoding outside the camera, they assume you will do single stage workflow or use an uncompressed workflow.

 

[Nityananda_SCSMath]

Native Editing of AVCCAM vs Transcoding to an Intermediate Codec - Quality Loss?

Native Editing of AVCCAM vs Transcoding to an Intermediate Codec - Quality Loss?

I don't know if this is the best thread for this conversation, but I think it would be good to clearly establish this point regarding the native AVCCAM codec of the camcorder.

The new NLE's Avid Media Composer 5 and Adobe Premier 5 both support in some manner native AVCHD/H.264 editing without transcoding. One BIG advantage of the AVCCAM codec is that the files are very small, making archiving AND editing the video require much less hard drive space. For those amateurs like myself who film long lectures, events, etc. in high def and then archive and store the footage with a small storage budget this is great.


I was thinking that if I had a fast enough computer for playback of timelines with just basic edits, crossfades, and some simple color correction that I would be able to edit the AVCCAM files natively and export compressed H.264 .mp4 files for web distribution, keeping my storage needs small and NOT losing any quality.



SO: Can you edit AVCCAM files natively in MC5/Premier5/etc. without losing any quality, compared to if you had transcoded into Prores and edited in Prores or a similar intermediate codec?
If so, at what point will you start to lose quality compared to editing in prores? (Color correction, multi-workstation workflow, etc. ?)


I greatly appreciate your assistance in making this point clear for everyone. I read the Steve Mullen Long-Gop article but I have to admit that I didn't fully get how it applied to AVCCAM.

Thanks for your help, looking forward to the AF100 !!!
-Chris

 

[KMR]

If you edit mpeg2 long GOP (HDV, XDCAM EX, DVCPRO HD and such) native

I do not believe that DVCPRO (any flavor) is either MPEG-2 or long GOP.

 

[Ben_B]

I do not believe that DVCPRO (any flavor) is either MPEG-2 or long GOP.

Yeah that statement sort of made me stop reading that post.

 

[Barry_Green]

The marketing of the all those "workflow" codecs is so strong that sometimes people forget that for a single edit cycle, the original codec is always better. Certain workflow codec manufacturer even support their sales with manipulated tests.

ABSOLUTELY AGREED. And boy, does that annoy me...

Junk->timeline processing->output codec is always better than
Junk->uncompressed->timeline processing->output codec and a lot better than
Junk->any workflow codec->timeline processing->output codec

Correct.

There could be other advantages though, the workflow codec could have lower cpu use. Putting that aside, the extra transcoding always reduces the output quality of a single stage workflow.

Absolutely correct.

Jan is probably talking about multistage professional workflows though, where processed material (not EDLs) is transfered from edit station to edit station. That requires multiple decoding/encoding cycles so you don't want to do that on a long GOP codec.

Agreed absolutely.

Now, here comes the caveat -- this all depends on the particular NLE being coded properly to work properly. What Otis is saying that I am agreeing with is that there is no reason why there should be ANY degradation due to cutting on a non-i-frame; NLEs can certainly be written to take that into account. But whether any particular NLE does it that way, or takes the lazy way out and degrades the footage on every cut, is up to that particular NLE.

 

[dvbrother]

The Broadcast Engineering Magazine article by Steve Mullen (see Nityananda_SCSMath's earlier post for the link) is an interesting read. I've been a Final Cut Pro user since version 1.0, but knowing that Adobe Premiere Pro CS5 edits AVCHD natively, and can also handle something like 13 streams of DVCPRO HD in real time, I may be checking it out. If Steve Mullen's facts are correct, then the re-encode to a non-long GOP format prior to editing is an unnecessary waste of time (provided your NLE can handle the long-GOP codec natively).
I currently edit XDCAM EX (from a JVC HM700) on Final Cut at my day job. It looks great. The point is, I don't think any undue recompression is going on when editing in these codecs.

 

[Barry_Green]

The new NLE's Avid Media Composer 5 and Adobe Premier 5 both support in some manner native AVCHD/H.264 editing without transcoding.

And so does EDIUS, and so does Vegas. In fact, most every NLE other than Apple's will allow it.

One BIG advantage of the AVCCAM codec is that the files are very small, making archiving AND editing the video require much less hard drive space.

Exactly. Converting to ProRes results in the file sizes inflating from 21 megabits up to as much as 220 megabits. If space is important, keeping the files in their native format would allow for much less disk space usage for transfer, during editing, and for archiving.

For those amateurs like myself who film long lectures, events, etc. in high def and then archive and store the footage with a small storage budget this is great.

Agreed. But it's also great for those who need to do a quick turnaround project. Same-day edits, or news editing, etc; being able to just immediately edit the raw files instead of having to process/convert those files into a different codec, can make a big difference in your overall delivery timeline if you're under pressure to deliver a quick turnaround.

I was thinking that if I had a fast enough computer for playback of timelines with just basic edits, crossfades, and some simple color correction that I would be able to edit the AVCCAM files natively and export compressed H.264 .mp4 files for web distribution, keeping my storage needs small and NOT losing any quality.

You are thinking correctly, with one possible issue -- you will always lose a little quality on your final render. Any time you recompress footage, the image is going to take a bit of a hit; how much of a hit depends on how you export the footage. But that's a given regardless of how you edit it, no matter what, so ... it's really not anything to worry about, I mean, even if you brought the footage in as uncompressed and edited uncompressed, it should be obvious that you'll take a hit on quality when exporting as h.264 (or as anything, other than uncompressed!) So I'm just clarifying for you that yes there will be a little bit of a quality hit on the final export, but that will happen regardless of what you edit as.

SO: Can you edit AVCCAM files natively in MC5/Premier5/etc. without losing any quality, compared to if you had transcoded into Prores and edited in Prores or a similar intermediate codec?

Yes.

If so, at what point will you start to lose quality compared to editing in prores? (Color correction, multi-workstation workflow, etc. ?)

Where you would start to lose quality is if you were rendering out files and then re-importing them and working on them again. For example, if you rendered out a timeline in h.264 to send it to After Effects, and then exported from After Effects in h.264 back to your NLE timeline, you'll take quality hits on each of those.

It is the re-renders that cause the quality hits. If you don't re-render, you won't ever encounter a quality hit.

 

[Barry_Green]

I do not believe that DVCPRO (any flavor) is either MPEG-2 or long GOP.

It isn't. All the DVCPRO variants, and AVC-Intra flavors, are all intra-frame only, and none are based on MPEG-2.

 

[Ben_B]

A lot of my footage goes out to After Effects or some program other than my NLE to get color corrected at some point, so I don't mind transcoding. Also storage is CHEAP, as long as the archive is small, I don't mind large files that run faster.

As for export, my stuff is either going on DVD, or Online, or being screened. If it's being screened we usually just run a massive ProRes file off a computer, and once it hits online or DVD any quality differences will be totally unnoticeable.

 

[Nityananda_SCSMath]

Wow. I posted my question 5 hours ago and now I have exactly the information I needed. Sweet. :thumbsup:

My interest in AVCHD/AVCCAM:
- start editing right away for a fast less-than-a-day turnaround (news, online subscribers)
- Minimize my hard drive requirements (mobile editing with small bus powered drive/laptop/mobile internet card and long-term archival with a nonprofit storage budget)
- Be able to record longer to my camera media, whether SDHC P2 etc.


In certain rushed situations I would be willing to sacrifice ultimate image quality to achieve these. The fact that editing in AVCHD all the way through to export is actually the best way to maintain image quality anyway just makes it a no-brainer.

I know AVC-Intra is a better quality codec but for me (mobile, fast, cheap internet news) AVCCAM is going to be just what I need. Now I just need to choose my NLE (get with it Apple! :-Mad(DBG)

From what I read Premier doesn't need to transcode for export to After Effects, etc. So I can stay in the original AVCCAM capture until I export, losing quality only once during the final export to a compressed consumer format.

Premier and AG-AF100 here I come..
Thanks for your help.
-Chris

 

[Otis Grapsas]

About the output codec loss Barry discusses, I don't think it should be counted as a loss. This loss will be there anyway, there is no alternative, we need to ouput to the compressed codec. The loss will be zero if one starts with uncompressed so we can't blame the output encoding for anything:

Uncompressed source: processing loss + output codec loss
Compressed source: compressed loss + processing loss + ouput codec loss

I would think all modern editors disregard intermediate renders when outputing a timeline and simply go the the source (input->processing->output). But I could be wrong. Clearing all temp files should force them to do that because they can save no time in this case (ever a programmer wouldn't degrade your footage if there is no performance gain). In this case the intermediate will not improve anything.

I doubt any editor performs a destructive cut. The editor just asks for the frame using the respective codec and gets it to feed the processing. There is no point in precutting the material, it would just waste time and storage and there is no performance advantage.

I recently saw a wedding videographer that cut all scenes from the original GOP codec files and imported the cut segments to the editor. This certainly adds a layer of GOP mess to the output and hasn't got any performance advantages over cutting segments non destructively from the original files in the editor. But this is an education issue.

One reason I started my camera project was to be able to get from the sensor to the final film output without getting through compression math of any form (no DCT, no wavelets, no transforms of any type) and to compressed forms of delivery by only suffering the necessary final output encoding transform. I even kept eveything RGB in order to avoid color space remapping, since sensors are naturally RGB. Then digital cinema (DCI) came to destroy my purist dreams by adding a layer of compression for the distribution even in cinemas

 

[Nityananda_SCSMath]

Right. We all understand that exporting your project to a 30MB H.264 file for vimeo is going to be a quality loss from the original! But I appreciate Barry's mentioning that step, because it makes the whole thing very clear.

Sensor Image -> compressed camera codec (AVCCAM) -> compressed end-viewer format (DVD, web, whatever.)


Everyone (including an AVC-Intra user) has to take the final hit - unless they publish their capture format, like somebody with an ARRI Alexa who theater-projects Prores from a computer!

So now we just have to decide if we are OK with taking the Sensor image and compressing it to Panasonic's AVCCAM. I know I'm OK with that, others may not be due to compression quality of AVCCAM vs AVC-Intra etc.


Just to confirm though, is it correct to say that other than compression image quality, AVCCAM has no disadvantage compared to AVC-Intra as an editing codec? I know it takes more processor power etc. to render/watch but I mean as far as editing goes.

Doesn't AVC-Intra also work like this?:

Where you would start to lose quality is if you were rendering out files and then re-importing them and working on them again. For example, if you rendered out a timeline in h.264 to send it to After Effects, and then exported from After Effects in h.264 back to your NLE timeline, you'll take quality hits on each of those.

It is the re-renders that cause the quality hits. If you don't re-render, you won't ever encounter a quality hit.