ATOMOS Subforum?

[David Elkins]

Originally Posted by James Kuhn

I just got the craziest email from Atomos announcing their attendance at the 2011

I got that email about a week ago. Strange indeed. Makes me wonder what is really going on there.

[Marty Hudzik]

Originally Posted by David Elkins

I got that email about a week ago. Strange indeed. Makes me wonder what is really going on there.

Let's not make too much out of this. I manage our company's Exchange server and have 14 years in IT industry and sometime odd things happen. To put it in perspective, my company has 7-8 programmers who write custom IOS and Android apps for clients. They are very competent and great programmers. They also have "absolutely nothing" at all to do with our email system thus, a rogue email sent out by our server (for whatever unknown reason) is no reflection on the quality of our projects. So....this should not be of great concern to us as end users.

Unless of course they have their programmers who write the Samurai GUI and code, administering their mail servers too.....in which case, they need more employees!

[Marty Hudzik]

Can anyone vouch for a cheap SSD that they have working with the Samurai? I have a couple of 500GB spinners for main storage but I'd like to get a small SSD for times where vibration could be an issue. I only need a 64GB to capture in those situations, so I am looking to keep this cost effective. Anyone using something that might fit the bill.....and it's actually working? I can just order one but I was hoping for someone who can actually vouch for a unit with the Samurai.

[James Kuhn]

Marty...'ghosts in the machine', eh? Believe me, I know how hard it is to bring a new item to market, as long as Atomos is taking care of the 'big things', I don't care about the small stuff.

Unfortunately, I can't offer you any suggestions for a 'cheap SSD'. I purchased an Intel X25, 160GB SSD prior to my Samurai arriving and paid a 'premium'. My opinion is, buy whatever you can get your hands on at this point. I don't think the 'market' has felt the full impact of the 'floods' in Thailand, I believe existing stocks are currently being sold and shortages due to manufacturing delays haven't hit us, yet. I hope I'm wrong.

Best regards,

J.

[Marty Hudzik]

I thought SSDs were not being affected by the flooding, at least not nearly to the level that Hard disks are. I could be wrong. So what SSD are you using with the Samurai James?

[James Kuhn]

Marty...I'm using an Intel 25X, 160GB, model: SSDSA2M160G2GC 2.5". I purchased the Intel SSD back on November 6, 2011. I also have (2) Western-Digital WD3200BEKT, 320GB 2.5" HDD, and (1) Western-Digital WD5000BPKT, 500GB 2.5 HDD. I can only vouch for the 320GB HDD at this point, as it's the only one I've used, so far. It loaded up the Atomos 'Firmware v 2.6' with know problem after 'formatting' in the Samurai and it seems to record and playback without a hitch. BTW, all of the HDD are 7,200k RPM. I tried to stay away from the 5,400k RPM drives to begin with.

Regards,

J.

[Alister Chapman]

Here's a couple of pretty convincing frame grabs from my Samurai codec stress tests. It's the same frame in both cases, on from the internal PMW-F3 codec and the other from Samurai ProRes HQ. The camera was shooting S-Log and the clips were both graded with exactly the same settings. While these are zoomed in crops, on a big TV artefacts like this make the entire scene look very "lively" and busy, when you get to 50" and above they become quite clearly visible.

[Marty Hudzik]

Alister,
Any chance you can post the full size frames? I'd like to see exactly how cropped these clips are.
Thanks.

[Alister Chapman]

Here's the video clips, PLEASE view full frame HD or download the original file from vimeo to really see the difference. Please consider the additional compression added by Vimeo.

PMW-F3 and Samurai Codec Tests on Vimeo

[James Kuhn]

Alister...thank you so much for the excellent 'A/B' test. I'm a 'neophyte' as it concerns DV, but it looked to me like the Samurai capture was a trifle sharper and less noisy. Am I 'seeing' incorrectly? I have a new Samurai, and at the risk of being burned at the stake for heresy, I purchased the Samurai for ease of work flow and it's 'bulk storage' capability, rather than it's image quality. Although, image quality was a close second.

Again, thank you.

Best regards,

J.

[Alister Chapman]

Exactly, the Samurai compared to most AVCHD and Mpeg 2 based camcorders can capture more picture information when you use the higher quality modes such as ProRes HQ.

Noise is an interesting one. At first glance the 35Mb/s original material can appear to have less noise as the codec must ignore or reduce fine detail and noise to squeeze everything into just 35Mb/s. But this process results in some of the blocking that you see in the zoomed in side by side clips. As noise is quite random and moves about these blocks appear to move around and this can look very ugly indeed. There is none of this in any of the footage that I have shot using ProRes HQ. The workflow with the Samurai is indeed about as simple as you can make it.

[James Kuhn]

Alister...thank you for your thoughtful response.

Regards,

J.

[Marty Hudzik]

James.....how dare you you!!!! Image Quality is second? Really???? :)

Seriously, the workflow is nice too, but I was going after the 4:2:2 and ProRes compression to update my Canon XLh1.

Okay, having said that. Allister....I have a question for you. I have not done side by side comparison of the HDV material that my camera encodes to, and the ProRes HQ, but.....it appears that I see more tiny grainy noise in my footage than I did with HDV. At least on my Computer monitor at full 1920x1080.

This had me dispointed initially but when I view the footage on my 46" LCD it looks immaculate and I cannot see that fine grain. It is not compression but almost like a dot pattern. It is very fine. Is this what you are talking about? Is this lost in the HDV compression along with other details?

I haven't viewed your footage yet...wil go grab it now.

Thanks.

[Alister Chapman]

HDV, like AVCHD (less than 60Mb/s) and Mpeg2 (less than 100Mb/s) Has to throw away a lot of very fine picture information to squeeze everything into the restricted bandwidth available, especially if there is a lot of motion in the image. Noisy cameras are the worst as the codec can't tell what is noise and what is real picture information, so a lot of both get lost. Uncompressed noise tends to look like very fine grain or perhaps sand dancing around within the picture.

The end result is that a cursory glance at comparison frames may appear to indicate less noise in the HDV/Mpeg/AVCHD than the Samurai or other similar device. In some respect this is true, there may be less fine grain noise in the HDV etc. But as well as less noise there will also be less real picture detail, textures can really suffer, things like fabrics, skin, grass. The noise and fine textures get translated into square blocks of picture information called macro blocks. Long GoP codecs like HDV/AVCHD work by predicting motion between frames, so as the original noise in the camera output dances about, the codec is often fooled into thinking the real image is moving and the macro blocks get moved about by the codec in response to the image noise. These dancing macro blocks are evident in the F3 clips I uploaded, especially on the concrete wall of the first clip. It looks ugly and with a noisy image they dance about a lot and as a result catch the viewers eye. In the hedge/house clip the macro blocks that make up the wall and hedge move in very slightly different directions and this makes the wall and hedge appear to ripple very slightly.

Often all these artefacts go un-noticed, that's how the codec is supposed to work, take a full frame image, discard as much as you can, in such a way that the viewer won't notice it. That makes the signal smaller and less data is required. This is fine if your only going one generation or not processing your image. But each time you copy or process the image and re-encode it, another layer of artefacts gets added in addition to the original ones and it's at this point that you start running into problems. Just consider a typical production workflow: Camera Encode: Transcode To Edit (essential for most AVCHD projects): Render Effects and Produce Master: Encode For Distribution. That's at least 4 layers of artefacts, one on top of another. This is where starting off with as good a copy of the cameras output as possible is essential if you don't want to see serious image degradation.

[Marty Hudzik]

Thanks Allister. That description helps me not "regret" my purchase of the Samurai. While my XLh1 is not of the same caliber as your F3, it is producing some might fine images. Still, when viewed on my high end PC monitor the "grain" pattern still seems objectionable to me. When viewed on my actual TV which is mcuh larger, the grain is not apparent, and the image looks sharper....so there must be image processing going on with the TV that the PC doesn't do. Regardless, it looks great.

I will say this.....even with your footage from the F3 going to the PRoRes HQ codec, the final images are not perfect. While it is still far better than HDV or other GOP compression, I can still see a lot of grain and noise in some areas of the images that I wasn't expecting. I think I was putting the ProRes codec up on a pedestal and expecting perfect imagery in every sense....kind of like the high res stills I take with my 7D. They are immaculate compared to video footage.

So I have to temper my expectations and realize there is still going to be some noise or grain in areas of the images when I view it on my PC. (the close up of the bricks by the water that you showed still had some noise even on the ProRes codec). This is probably true of properly shot "35mm" also.....I just don't view that material with the same eye that I view my own clips...plus I don't view it on my PC....only my TV which obviously displays a better picture for final viewing.

Thanks again. Nice looking footage too. The F3 looks nice. I'm jealous. :)