Mathematically lossless - meaning it can really do the job in multi-generation work.

Another test compared to SONY Mpeg2 in MXF container which can be a good source for DVDAPro reveals that this MXF codec maybe slightly better than XAVC of the same SAR and PAR - but while very small in comparison to MagicYUV - the losses suffered by SONY MXF ARE so utterly revealed.

1) It would be more likely to gain traction if it didn't have the word "magic" in its name.
2) SMPTE registration ?

Bob.

There is a lot to be said for choosing a name and creating a brand. Do you recall the original developers of MPEG2 video encoding - their company was called Ligos - I was aboard their research team years and years ago. Who remembers? The fate of inventions means that someone else will eventually re-brand it.

But I digress, AND now more about Magic. Actually it is a bit humorous, because the codec gives exactly that feel of magic, so the name applies. But, branding is a special topic and this one is most curious.

For those with good memory banks, you will recall several efforts over 10 years ago to mainstream open-source codecs and one was branded Magic. It did not get very far.

Note this is a different party to the table as far as I can tell.

Is the name MagicYUV really any weirder than "Lagarith"?

I haven't looked into the possibilities, but so far I got one of these.

don't understand what John Dennis is illustrating. When you installed - did you get any error messages at all about compatibility with your hardware? The development cycle for this product has been smooth over the last few months - but admittedly it takes a large beta phase to shake out all the parameters. As more users attempt it there should be more to say,

For myself - it is buttery smooth and not one glitch yet.

My apologies.

I looked at the Mediainfo report for the failing render and found that I must have mistakenly selected the Sony YUV codec.

Complete name : C:\Users\John\Desktop\MKZ.avi
Format : AVI
Format/Info : Audio Video Interleave
Format profile : OpenDML
File size : 3.72 GiB
Duration : 40s 40ms
Overall bit rate : 797 Mbps
Encoded by : John Dennis
TCOD : 0
TCDO : 400400000

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 : 40s 40ms
Bit rate : 795 Mbps
Width : 1 920 pixels
Height : 1 080 pixels
Display aspect ratio : 16:9
Frame rate : 23.976 fps
Color space : YUV
Chroma subsampling : 4:2:2
Compression mode : Lossless
Bits/(Pixel*Frame) : 16.000
Stream size : 3.71 GiB (100%)

A more recent MagicYUV render looks like this:

Complete name : C:\Users\John\Desktop\MKZ 4.avi
Format : AVI
Format/Info : Audio Video Interleave
Format profile : OpenDML
File size : 1.92 GiB
Duration : 40s 40ms
Overall bit rate : 411 Mbps
Encoded by : John Dennis
TCOD : 0
TCDO : 400400000

Video
ID : 0
Format : MAGY
Codec ID : MAGY
Duration : 40s 40ms
Bit rate : 409 Mbps
Width : 1 920 pixels
Height : 1 080 pixels
Display aspect ratio : 16:9
Frame rate : 23.976 fps
Bits/(Pixel*Frame) : 8.234
Stream size : 1.91 GiB (100%)

Lower the Magic YUV glitch count to 0.

Raise the Sony YUV glitch count to 1.

On the strength of a little testing, this codec really is magic!

I've been using UT Video Codec for a few years, now up to version 14.2.0, and I've been very happy with it. I prefer it to HuffYUV and Lagarith because the playback is faster if "Optimize for decoding speed (predict left)" is selected in the options.

Now I have tried MagicYUV version 1.0 and basically it's better in every way. Thanks videoITguy! Unless I run into problems, I think this will be my intermediate of choice now.

Here are my quick test results. I tested using VP13 on my i7-5960X machine with Windows 8.1. 1 x AMD HD6970 and 1 x NVIDIA GTX580. I left all the MagicYUV settings at their defaults. Playback was at Best (Full). The playback figures are average rates. The actual output between the 2 codecs is identical to uncompressed.

[Edit: I later added the Grass Valley lossless codec and Grass Valley HQX codec at "Online (SuperFine)" setting to see how a near-lossless codec compared.]

TEST 1

Panasonic GH4 UHD 3840x2160 30p > Vegas Pro render > Playback with color curves & secondary color correction

                            Render   |  File   |  Bit     |  CPU     |  AMD   | NVIDIA

                             Time    |  Size   |  Rate    |  Play    |  Play  | Play

-------------------------------------------------------------------------------------------

8-BIT ORIGINAL

GH4 UHD MP4                   ---      0.28GB      81Mbps    9.0 fps   30 fps   29 fps



8-BIT LOSSLESS

UT Video Codec RGB          71 secs    6.34GB    1816Mbps    6.8 fps   22 fps   20 fps

MagicYUV                    42 secs    5.17GB    1481Mbps   10.5 fps   25 fps   23 fps

Grass Valley lossless      116 secs    9.56GB    2764Mbps    5.3 fps    8 fps    8 fps



10-BIT LOSSLESS

Sony 10-bit YUV*           108 secs   18.50GB    5303Mbps    6.1 fps   11 fps   11 fps



8-BIT LOSSY

Grass Valley HQX SuperFine  93 secs    3.67GB    1052Mbps    6.5 fps   14 fps   14 fps



10-BIT LOSSY

Sony XAVC Intra             65 secs    0.96GB     267Mbps    8.0 fps   30 fps   27 fps



*Some very slight movement of a few pixels when comparing on the waveform

TEST 2

HDV 50i > Frameserver > AviSynth QTGMC & resize to 1920x1080 50p > VirtualDub render > Playback in a 25p project with color curves & secondary color correction & Film Convert

                    Render Time | File Size | CPU Playback | AMD Playback | NVIDIA Playback

-------------------------------------------------------------------------------------------

UT Video Codec RGB   112 secs      3.68GB         9 fps          9 fps         12 fps

MagicYUV              77 secs      3.17GB        13 fps         20 fps         18 fps

Playback of either codec with mulitple FX in Vegas Pro 13 is not very stable. I get occasional crashes just while playing back. I'm yet to establish a pattern of whether that is happening more with one codec than the other, or with GPU more than without GPU.

Playback in VP12 was slower in all cases but it didn't crash. VP12 is more stable than VP13 for me.

My tests of MagicYUV in the standard mode alongside all of the popular intermediates like Sony YUV, MXF, Avid, and Cineform - show it to be a very stable codec and better than all the others. I 've tested XP32bit OS as well as VegasPro13 on Win7 64bit.

I think stability may be a factor based on the platform and hardware - that is where this developing software needs a lot of worldwide testing.

The public release of MagicYUV does not support 10 bit.
Cineform, Sony YUV, Dnxhd, Pro Res all support 10 bit.

MagicYUV is currently closed-source, unregistered, unlicensed (by the author's neglect, I imagine).

Apparently he has added 10-bit support to MagicYUV but only in private builds. Maybe he's planning to charge for that version, or maybe he just wants to get more testing done on it before it's public.

Since release 14.0.0 UtVideo now has a Pro YUV422 10bit option. I haven't tried it.

The later versions of Helix codec support 12 bit and 8K resolution.
They're free, but download them because the versions in Vegas may be older.
http://forum.doom9.org/showthread.php?s=&threadid=56972

Nick,

Did you test Cineform, and how do it's render times and playback compare?

Also, did you do any video quality comparisons? You mentioned Lagarith and HuffYUV which are lossless and therefore, theoretically, should not change or compromise quality. However, the OP mentioned he used Cineform, as do I, and it can make very subtle changes in quality, depending on the settings chosen, although those changes are usually virtually imperceptible.

The SDK kit for 4K magicyuv is available - this is how the program will generate its only income.

I test against the intermediates (one of my favorites for all time is Cineform and you all knew that already). I would use Cineform any day of the week, but the stress test for Magicyuv is maintaining integrity in 32bit video mode in VegasPro and bests Cineform in that single comparison. So there is a particular high-value high-end reach for the MagicYUV even in the 8bit codec.

@VideoITguy - Are you saying that MagicYUV bests Cineform in image quality? Is it also 10 bit? I see the tag line on the site saying 4K 10 bit but does that translate over to HD footage as well? Cineform has been my go to intermediate codec but to get anything more you gotta pay to play - and given that the latest version of Premiere Pro CC reads Cineform natively instead of needing the various DLL files installed by their apps, I wonder if there's a catch. I tend to play conservative when it comes to post production tools (as my recent postings on still sitting on the fence with Vegas Pro has shown) so I'm somewhat hesitant to try something else - especially given there's no real technical support either free or paid to resolve issues quickly in a production environment. Maybe you know something else I don't and I"m all for moving to something better if it's more or less ready for serious use.

Is it also 10 bit?

From

I am pleased to announce that MagicYUV now supports encoding and decoding of deep-color (10 bit) video material, as well as the availability of the MagicYUV SDK.
Deep-color support is not included in the public releases of the codec. If you would like to have more information about it or would like to request an evaluation license, contact me.
If you would like to learn what kind of 10 bit formats are supported go to the features page.
To read more about the SDK click here.

Mike

MagicYUV is an engineered lossless codec approximating uncompressed - so in a lot of ways immune to the compromises of compression codecs like Cineform.
I have not a clue what all of the 4K SDK kit includes - it might require engineering dlls but I doubt it. The free codec is 8bit and that is all that VegasPro can really deal with on its own without dlls -but you can work it into a 32bit workflow - better than almost any other codec I am aware of. Now the 4k SDK is set to 10bit - but keep in mind that's a lot larger platform than you and I deal with. I would discourage your thinking that Premiere 10bit handling is anything but about the same as what you will push out of VegasPro.

That reads like an advertisement.
In truth, there are several YUV VFW codecs that can be used in Vegas, they all use flavors of Huffman encoding, and their lossiness is largely a byproduct of chroma subsampling, not from any "magic."

Because they're all huge, their appeal is somewhat limited. Here is a partial list of ones I've already tested in Vegas, with similar results.

Sony Yuv
Huffyuv
Lagarith
UT
Helix
Canopus
Matrox

No one has said its magic, and there is no implication its magic. Quite contrary its coding set to a difficult parameter to meet.
Increase the speed of rendering, while downsizing the output file size while at the same just keeping the highest quality, then add stability, and versatility across platforms.

when you guys figure this out, can you make me a brief idiot install sheet so I can get the best 4k possible? thanks!

Nick, Did you test Cineform, and how do it's render times and playback compare?

John, I'm afraid I didn't test Cineform. Only these 2 lossless codecs. I've only used UT Video Codec for intermediates in the last few years and it's been very good to me so I don't see the point to downgrade to a more lossy codec for working on material that originated as HDV. I have plenty of disc space and playback is smooth enough.

Also, did you do any video quality comparisons?

MagicYUV at default settings and UtVideo Codec RGB were truly lossless. Not a single pixel moved in the waveform or RGB parade compared to uncompressed. However there was a slight shift when I tested UtVideo Codec YUV422 709.

So I'm happy so far to change to MagicYUV for work originating in 8-bit formats. If I wanted a near-lossless 8-bit codec now in Vegas Pro to save space or whatever then I would look seriously at the free Canopus AVI codecs. If they're good for Malowz they're probably good for me. DNxHD has the Quicktime drawback, and there have been so many versions and licensing issues with Cineform over the years that I'm just left totally confused by it.

I'm hoping to get an Atomos Shogun when it's out and record in 10-bit 4:2:2 via my GH4. Then I might occasionally need a 10-bit lossless or near-lossless intermediate from another codec, but really I'm hoping to just work on the native files.

Thanks Nick!