F3 4:4:4 recording solution? Forget the gemini!

[Peter G. Johnson]

Attached are two images. They are still frames of video at 800% magnification and saved as PNG files. One is using the ProRes422 10bit 220mbps (Pix 240) and the other is an identical time from the SxS card 8bit 35mbps.

[Nate Weaver]

Holy cow. I knew XDCAM was a little softer, but not like that.

[Tom Bostick]

Originally Posted by Peter G. Johnson

Attached are two images. They are still frames of video at 800% magnification and saved as PNG files. One is using the ProRes422 10bit 220mbps (Pix 240) and the other is an identical time from the SxS card 8bit 35mbps.

the difference in those is huge!

[Peter G. Johnson]

By the way, those are not wrinkles. I was squinting as much as possible to create more detail for s-log.

Originally Posted by Nate Weaver

Holy cow. I knew XDCAM was a little softer, but not like that.

Remember, that's 8x the original size. Most images look quite ordinary when you magnify them like that.

[Nate Weaver]

Originally Posted by Peter G. Johnson

Remember, that's 8x the original size. Most images look quite ordinary when you magnify them like that.

Oh, I caught that. You can't convince me it's still not a big deal. When you manipulate the image later in color or otherwise, those 'softer' areas fall apart before everything else.

I understand this might seem like splitting hairs, but after shooting Red for 2-3 years, I'm excited to see this cam close the gap.

Also, kudos for the 'angst' codec test. Tektronix might be interested!

[Alister Chapman]

I've got Samurai, Gemini, Gemini transcoded to ProRes and Native 35Mb/s frame grabs on my site for download:

DPX, ProRes and EX 35Mb/s comparison tests – what a nightmare! | XDCAM-USER.COM

Gemini is the clear IQ winner in this test. Interesting ProRes results, judge for yourselves.

[Steve Kimmel]

Thanks Alister. I tried to download files (put in my email), but it didn't seem to work.

[Alister Chapman]

The link should be working now. Any further problems let me know. Thanks for pointing it out.

[Steve Kimmel]

Originally Posted by Alister Chapman

The link should be working now. Any further problems let me know. Thanks for pointing it out.

Thanks, that worked! I see what you mean in your review on your blog.

[Cees van Kempen]

Originally Posted by Peter G. Johnson

Attached are two images. They are still frames of video at 800% magnification and saved as PNG files. One is using the ProRes422 10bit 220mbps (Pix 240) and the other is an identical time from the SxS card 8bit 35mbps.

Is the difference related to ProRess vs. XDCam or to 10bit 220mbps vs 8bit 35mbps? Or in other words, will capturing with nanoflash at XD Cam HD422 @ 220mbps be comparable with the proress example or with the SxS card example? I don't have a F3 (yet), but have a nano and consider the F3.

[Dan Keaton]

Dear Cees,

I would like to comment on your question.

Background: My image quality information is based on many emails from Sony ICE Team members who have an F3 and a nanoFlash. ICE = Independent Certified Experts.

What I hear most often, is that the images are "just stunning".

Here is why I believe this combination creates great images.

1. The Sony F3 is an exceptionally low noise camera.

2. The nanoFlash with the Sony XDCam 422 codec, running at higher bit rates is an exceptionally low noise codec.

3. Under normal circumstances, a codec has to compress the image which includes the inherent camera noise.

The noise makes compressing the image much harder as the noise is just more detail to be compressed to the codec.

But, when the image to be compressed, and there is very little noise, then the codec is not hampered by having to compress all of the extra detail caused by the noise.

Thus, the Sony F3 + nanoFlash can produce outstanding images. All of the compression "Horsepower" can be devoted to the image, and very little if any has to be devoted to the noise.


Alister has mentioned that the ProRes codec adds some noise.

Others have expressed this same opinion, based on tests that they have run.


It is very reasonable to record the images from the F3, with a nanoFlash.
Cinegamma 1, for outdoors or bright highlights and
Cinegamma 4, for darker conditions may be used.
(Your choices may vary.)

For the very best image quality, and for full support for Sony S-Log, we recommend the Gemini 4:4:4.

[Alister Chapman]

I shot some footage on the Gemini, internally on the F3 and and some SR-SQ using an SR-R1 today. S-Log and standard gamma. I'm on the road flying home tonight but as soon as I'm home I'll pull off some frame grabs. Really interested to see how the SR compares to the Gemini. I think I know which will look best, but how big will the difference be?

[Duke Marsh]

Originally Posted by Dan Keaton

It is very reasonable to record the images from the F3, with a nanoFlash.

Dan, I loved my Nano when I had it. Compared to the latest crop of recorders it draws negligible juice and minimal drive space. It's reliable, compact and light.

I know it can't, but I really wished it had been upgradable to 10 bit. If the Nano had 10bit I'd be on it in a minute. I know 10 bit isn't the be all and end all. As it is its still a good match for the F3. The Gemini was just a little be too much for my work. I'm in between your two products right now and not 100% happy with what I have. LOL

[Dan Keaton]

Dear Duke,

I understand.

One point to consider, our Sony Codec is lower in noise than the other recorders.

This does help the nanoFlash compete very well against the others.

For true 10 bit, with zero noise, the Gemini 4:4:4 is the answer.

[Alister Chapman]

I've been comparing ProRes to a number of codecs and the noise ProRes generates. This noise acts can become the limiting factor when you try to grade it. As an example I recorded some ProRes HQ and Mpeg2 100Mb/s on a NanoFlash from the F3. I find I can grade both by similar amounts before getting unacceptable artefacts. DPX from the Gemini, SR-SQ or DNxHD are much better.