The AMT.xsd in SDK 2.8 only lists 145 and 220 as valid bitrates, yet the XML examples included use 36, 100, 145, 220 and 440. The tag name is also stated as 'mBitsPerSecond' in the xsd but dnxhdMbitsPerSecond in the XML examples. I assume the latter is correct.

If I wanted to make DNX185 what would I put as the bitrate? Likewise 175, 115, 110, 90, 60 and 45. Are these formats not supported? I tried to make 720p/29.95 but 75 or 110 were not accepted for the bitrate.

A complete list of what bitrates are valid would be useful. The xsd only covers 1080i59.94 formats I'm guessing, and it doesn't appear from the above test that the full range of DNX formats listed in the Avid DNxHD whitepaper are currently available.

Thanks
Nick

I'm also interested in getting a list of supported codecs in AMT 2.9. 

Does anyone have a release note doc that lists the current formats in 2.9?

Thank you,
Matt Schneider 

Matt,

the list of supported codecs is in the AMT documentation. Please remember that AMT still needs to be licensed; as AMT licensee you should have access to the documentation you are looking for.

Thanks Andy. But it seems the White Paper on the Avid web site lists codecs from AMT 1.6:

http://www.avid.com/us/partners/AvidMediaToolkitProgram

I work for Technicolor-PostWorks in New York and we are not an AMT licensee. However we rely heavily on 3rd party support of Avid codecs in tools like Colofront OSD and BMD Resolve.

How may I obtain documentation on AMT? Please advise. Thank you!

Matt

I am an AMT licensee and I'm still not sure that there is a clear list of what DNX formats are supported in the SDK documentation. If anyone can tell me where to look please do so.

Thanks

Nick

Hi all,

unfortunately, there is no simple but exhaustive answer to the AMT format questions. AMT has two different modes, pass-through and transcode. Pass-through is the more straightforward case to describe and what this posting is about; for transcoding we would have to deal with the added challenge of allowable transcoding paths and source/target combinations which exceeds the scope of what can be documented in this forum. Even for pass-through, the format variety is extensive.

In pass-through mode, AMT 2.9 supports the following DNxHD compression classes through the following selectors:

Selector 220 (HQ - High Quality):

- DNxHD HQX 1080p (10bit progressive): DNxHD 175x, 185x, 220x, 365x, 440x
- DNxHD HQ 1080p (8 bit progressive): DNxHD 175, 185, 220, 365, 440
- DNxHD HQX 1080i (10bit interlaced): DNxHD 185x, 220x
- DNxHD HQ 1080i (8 bit interlaced): DNxHD 185, 220
- DNxHD HQX 720p (10 bit progressive): DNxHD 90x, 110x
- DNxHD HQ 720p (8 bit progressive): DNxHD 90, 110

Selector 145 (SQ - Standard Quality):

- DNxHD SQ 1080p (progressive): DNxHD 115, 120, 145, 240, 290
- DNxHD SQ 1080i (interlaced): DNxHD 120, 145
- DNxHD SQ 720p (progressive): DNxHD 60, 75, 120, 145

Selector 100 (CB - Compact/TR - Thin Raster):

- DNxHD CB 1080p (progressive): DNxHD 80, 85, 100
- DNxHD CB 1080i (interlaced): DNxHD 85, 100
- DNxHD CB 720p (progressive): DNxHD 40, 45, 50, 85, 100
- DNxHD TR 1080i (interlaced): DNxHD-TR 120, 145

Selector 36 (LB - Low Bandwidth):

- DNxHD LB 1080p (progressive): DNxHD 36, 45, 75, 90

This explains why for example the dnxhd_1080p_185_25 specification looks like this (i.e., contains the '220' selector), or in other words dnxhdMbitsPerSecond should be read as something 'dnxhdMbitsPerSecondClass':

<?xml version="1.0" encoding="ISO-8859-1" ?>
<AmtMediaAttributes>
    <format>
        <mediaFormat>Compressed</mediaFormat>
        <signalStandard>SS_274M_P</signalStandard>
        <dropFrame>kAVXFalse</dropFrame>
        <compressionType>Avid::HD::DNxHD</compressionType>
        <dnxhdMbitsPerSecond>220</dnxhdMbitsPerSecond>
        <dnxhdComponentBits>8</dnxhdComponentBits>
    </format>
    <timecode>01:00:00:00</timecode>
    <bitRateMode>CONSTANT BIT RATE</bitRateMode>
    <framesPerSecond>25</framesPerSecond>
    <storedHeight>1080</storedHeight>
    <storedWidth>1920</storedWidth>
    <audioTracks>2</audioTracks>
    <audioBits>16</audioBits>
    <audioBitRate>48000</audioBitRate>
    <audioCompressionType>Avid::WAVE::Uncompressed</audioCompressionType>
    <extendedAttributes>
    </extendedAttributes>
</AmtMediaAttributes>

Once a compression class has been selected, the exact target format is then a function of the remaining parameters (specifically, the frame rate) in the format specification.

Be aware that labels might change in the future as the naming scheme is further streamlined and standardized. The AMT documentation will be amended to contain a more detailed version of this overview; until then the XML sample files and the XSD schema are your best source of information.

Hope this helps.

Andy

Hi Matt,

for all API requests, please contact our 3rd party relations team at

[email protected].

They have a template that you can use to request an AMT license.

Rgds,

Andy

Hi Andy,

Many thanks for your reply, and for the information. "Passthrough" sounds like the AMT equivalent of "consolidate" in Media Composer terms (meaning the source codec is preserved and not recompressed) and transcode is transcode. 

Looking ahead, there are a few codecs of significance that are absent, and would like your take on what we can expect for the near future of AMT:

Canon XF (which is MPEG-2 50Mb 422); I suppose transcoding to XDCAM-HD (also MPEG-2 50Mb 422) would work, but I imagine this would be a transcode and not a pass-through, whcih would represent a generational loss in picture quality. 

And of course, there is ProRes wrapped in OP-Atom MXF. Working in ProRes in Media Composer is gaining more and more acceptance: is there a plan to include pass-through of ProRes wrapped in OP-Atom MXF in future releases of AMT? Media Composer does this now via consolidate on Mac and Windows, and transcode on Mac. So if Media Composer can do it, is it fair to assume AMT is not far?

Also: other requests commonly heard through out the user base is the introduction of audio sample rate addressing that permit sub-frame slipping in "film projects" (identifying the perf so one can slip based on perf), and the need for DNxHD 80. What is the future for this as well?

Thank you again. 

Matt Schneider - PostWorks New York

Hi Matt,

with the topics you raised, I thought it would be worthwhile to summarize a few general informations about AMT that may not be obvious at the first glance:

- AMT is a Software Development Kit (SDK) with the main functionality of reading and writing basic native Avid media, ie Avid-flavored OP-Atom MXF containers containing many of the essence encodings that are natively supported in  Avid products. It does not have the ability to directly read or write non-Atom containers such as MXF OP-1a or QuickTime(TM)  except for an intermediate format called 'AMT clips'. AMT is a frame-based interface and as such produces and consumes picture and audio sample buffers.

- To use it, integrators need to build and test their own applications for Windows or Mac. Since AMT primarily provides the interface to the Avid media world, this typically requires integrators to use additional SDKs such as a commercial MXF toolkit or the QuickTime SDK for accessing non-Atom media such as MXF OP-1a or QuickTime.

- AMT supports master clips only, in that sense 'pass-through' is not really equivalent to the consolidate function in Avid editors. Pass-through mode simply wraps (or unwraps) raw frames to and from OP-Atom. This is a lossless operation. It does not deal with sequences; those must be mixed down to master clips before they can be processed in AMT. In that sense, the difference between AMT and a full featured AAF media engine such as the player component in Media Composer is massive.

- My posting above focused on the VC-3 compliant DNxHD codec family. AMT can handle many more essence encodings and in many cases, can transcode between them. Supported encodings include 1:1 YCrCb and RGB 4:4:4, AVC-Intra 50 and 100, DNxHD, DVCAM, DVCPRO 25 and 50, DVPROHD, H.264 800 kbps, 1500 kbps and 2 Mbps, IMX 30, IMX 40, IMX 50, JFIF 1:1, 2:1, 2:1s, 2:1p, 3:1, 3:1p, 3:1m, 4:1p, 4:1m, 8:1m, 10:1, 10:1m, 14:1, 15:1s, 20:1, 28:1 and 35:1, XAVC-Intra 50 and 100 as well as XDCAM HD 422 50 Mbps, XDCAM HD 18, 25 and 35 and XDCAM EX 35 plus a few less frequently used formats.

(Note: This list is a high-level view. Not all combinations of codec, raster and frame rate are supported. For more details on the supported codec, raster and frame rate combinations please see AMT Functions Reference Guide, Section 2)

- Canon XF is from my perspective a meta format that uses MXF OP-1a containers and XDCAM HD essence encoding as you stated. As mentioned above, AMT focuses on reading and writing Avid OP-Atom; implementing support for other formats (such as OP-1a or Canon XF) would be a responsibility of the integrator. Note that the Avid AMA technology supports 'volume' meta formats such as XF but AMT 2.9 does not support AMA plugins.

AMT includes basic perf slipping support but perf slip has specific constraints regarding the editor setup. If you have specific comments please submit them through [email protected]. DNxHD 80 should be supported for 23.98p and 24p frame rates through the 'Class 100' selector.

You might be interested in a recently released Avid product called Media | Director that uses a service component built on AMT. The Media | Director service component implements some of the features you seem to be looking for (such as AMA plugin support) and adds an orchestration engine and a Media Central | UX based GUI.

Andy

(Note I made some small edits to this post)

Thank you Andy - that is all very helpful and appreciated. 

I am predominantly interested in 3rd party systems such as Colorfront OSD, Blackmagic Resolve, and other color systems including FilmLight Baselight, Nucoda Film Master, etc being able to create OP-Atom Avid media, using their own softrware as the front end. 

For example, a dailies-color software such as Resolve could create ProRes wrapped in OP-Atom just in the same way that Media Composer can currently do this, but you have the added benefits of applying a grade, LUTs and CDLs on a color system designed for that kind of workflow. Yes Media Composer can do some of these things, but isn't exactly wired for grading and color management the same way of these color-specific tools are. 

So indeed, my expectation is that the host application - OSD, Resolve, MTI Cortex, Nucoda etc - is decoding camera original formats up front, and is able to encode to OP-Atom MXF wrapped codecs that are Avid friendly as a deliverable. 

To this effect, sub frame slipping is still something that you cannot do even if you are in a "Film Project" in Media Composer. "Audio Sample Rate Addressing" is the typical error, but it means that the audio was created in an application using AMT and not Media Composer natively. 

Thanks again Andy. 

Matt Schneider