- April 25, 2020 at 7:53 pm
so far I’ve often used the Same as source option when exporting from Avid without any problems, even if the timeline contained various codecs (but always DNxHD), for example DNxHD 36, DNxHD120 and DNxHD185x. It resultet in faster export and produces the same result as if I exported it using Custom Settings to export to DNxHD 185x.
Untill now I thought that Same as source export used the highest resolution that was used in the timeline and exported it in that resolution, while transcoding other resolutions to the highest resolution. But I recently read somewhere that it actually produces a file that contains multiple codecs (DNxHD 36, DNxHD 120 and DNxHD 185x), while being laballed as DNxHD 185x.
It was a very old post, but I’m kind of worried now and I’m wondering whether it is true. So, how does Same as Source export actually work?
- April 25, 2020 at 8:12 pm
If you have more than one codec type on the timeline, do a VIDEO MIXDOWN first, and then do an SaS export.
Now, as for the mixed codec thing, no, that’s impossible. You cannot have a video file that has starts as one codec, switches to another, and then another. Things don’t work that way. BUT, not having done an export with multiple types myself (always mixed down), the way to see which one it is, is to re-import that file and see what the information in the bin says.
Little Frog Post
Read my blog, Little Frog in High Def
- April 25, 2020 at 9:30 pm
Hi Shane, thank you for the reply!
In the future I most definitely will use a video mixdown. The thing is that usually when I’m exporting from Avid I receive the graded file, that doesn’t have any cuts anymore and I just add the opening credits, final credits and perhaps some FX that was rendered elsewhere. Those are usually the elements that are not DNxHD 185x and represent a little percentage of entire timeline.
Would you say that exporting the timeline as “Custom” to DNxHD 185x without doing a mixdown would yield the same (or really similar, without noticable differences) result as doing a video mixdown and exporting it “Same as source” in the specific situation that I have described?
I now also did a test: I imported the same clip with a duration of around 5 seconds as DNxHD 120 and DnxHD 185x.
– I tried exporting both of the clips on timeline as Same as source and the result is strange. Since each codec represents half of the timeline, Avid exports the clip with a different bitrate: in my case 165 Mb/s. MediaInfo doesn’t label it as anything specific but it does say it is a DNxHD codec by Avid. When I import it back to Avid it just says “Avid DnxHD Codec”.
– When I exported a whole clip as Same as source that was DNxHD 185x and just a few frames of the DNxHD 120 file I got a different result: MediaInfo labelled the exported clip as DNxHD 185x by Avid and the bitrate was in fact 185 Mb/s. However when I brought it to Avid with Source Browser it again only said “Avid DNxHD Codec”.
– Exporting only the portion of the timeline where the DNxHD 185x clip was resulted in an expected outcome: MediaInfo labelled it as DNxHD 185x by Avid and when I brought the clip back to Avid it was labelled as DNxHD 185x.
So I guess that if you use the Same as Source export option Avid will always export one ONE sort of codec with a constant bitrate. In the case of mixed codecs it will find a middle way and export it with a non-standard bitrate, but if one of the codecs presents the majority of the timeline it will label it as that particular codec.
I’ve been an editor for a while and I now believe it was definitely not ok that I sometimes didn’t use a video mixdown. But would you say that if the exported files were labelled as DNxHD 185x and I had no issues with them that those files are still ok in a technical sense? I’ll be more careful in the future for sure. 🙂
- April 26, 2020 at 7:32 am
If you have a mixed codec timeline and mix down as SAS you will get a file with mixed codes. The metadata of the file will be flagged as the first codec used.
This normally isn’t an issue as most players can switch codec mid playback but not always. So it’s best to mix down or transcode first.
Selecting custom codec and then choosing a DNX codec in the QT option is a real no no due to gamma shift issues.
- April 26, 2020 at 10:43 am
Thank you for the reply Pat.
I’m a bit confused now, because I got 2 different answers about the question about a file that is exported with different codecs on timeline as Same as Source. Shane said it was impossible for an exported file to contain multiple codecs, while you say that this is the way Avid SAS export works. Do you perhaps have an external link in which this is verified for sure?
“Selecting custom codec and then choosing a DNX codec in the QT option is a real no no due to gamma shift issues.”
Could you please elaborate on that? Are you saying that if you have for example a clip with Prores 4444 (that was brought in with source browser) and Custom export it to DNxHD 185x it will result in gamma shift? I just recently had that situation and when I compared the exported file in to the original Prores grade in Avid Media Composer, Davinci Resolve and VLC Player I didn’t notice the any gamma shift, the only difference was slight compression that was visible on the Y waveform but nut with the bare eyes.
Do you think that Custom export as DNxHD 185x should never be used if the gamma shift is a posibility? Could the gamma shift also happen with a clip in timeline that was transcoded to DNxHD 185x in the import process and is them Custom exported to DNxHD 185x?
- April 26, 2020 at 3:24 pm
I’m more worried that you are using dnx 36 material on a timeline, and exporting it as 185x, thinking you are gaining some advantage with that footage.
Typically Dnx 36 should not be part of your final export.
- April 26, 2020 at 3:29 pm
The gamma shift issue is a well know historic limitation of the, QT export engine. It doesn’t always happen but its problematic enough that serious post houses just don’t use that route.
Ensuring all you timeline media is at or above your finishing quality.
No DNx36! Then with transcode the entire sequence. Or mix it down to finishing quality then SAS export.
- April 26, 2020 at 5:12 pm
Thank you both for replying.
The DNxHD 36 part was plain white titles on black bacground that I received from another person using another software. Since I presumed it wouldn’t make that much of a difference if it were exported at much higher bitrate, because there were only white black and some grey pixels I didn’t ask for a better export. Will do in the future however! I am perfectly aware that exporting something of a lower quality to a file with higher quality will not improve its quality.
I have heard of this issue before, but when I decided to export with custom settings for practical reasons, I always checked the exports in different softwares and didn’t notice any significant change besides compression.
So, the best option for final export for the archive of a project would definitely be to do a video mixdown of an entire sequence to a wished DNxHD or DNxHR codec and then export Same as Source, which would NOT result in a gamma shift? Would exporting in MXF OP1A with the wished DNxHD or DNxHR codec also be a good or perhaps better option?
And if you perhaps also know about the exports from Premiere Pro and Davinci Resolve? Is Quicktime DNxHD/DNxHR also sometimes problematic from those softwares in terms of gamma shift? If that is the case, what would produce the same wished result? MXF OP1A DNxHD/DNxHR? Prores is not an option for me most of the time, because I do the wast majority of my editing on Windows computers.
Thank you for taking the time for replying. Really appreciate it!
- April 26, 2020 at 5:21 pm
MXF OP1A is fine for export and a good choice. Any software that uses (or Used) the legacy QT for its export was prone to the gamma shift.
DNx36 should be fine for a B&W title (and you would be amazed how much DNx36 makes it to air)
- April 26, 2020 at 5:36 pm
Interesting. But Same as Source option doesn’t use the legacy QT for its export and is therefore not dangereous to produce gamma shift?
Log in to reply.