EX1R - HDMI output 8bit or 10bit?

[Cliff Totten]

Can anyone confirm the bit depth of the EX1R's HDMI port?

We know that it's SDI is 10bit but can we assume that it's HDMI output is the same?

also, I could very easily be wrong about this but I was told by several people that the general HDMI spec for all video devices only supports an 8bit color space.

I'm looking into getting an Atomos Ninja and I'm hoping I can grab true 10bit video of my EX1R's HDMI port.

CT

[Olof Ekbergh]

8 bit HDMI and HDSDI. It is an 8 bit camera.

But the color space is 422 on HDSDI and HDMI.

The EX codec is 420.

[Alister Chapman]

That's not correct Olof.

The HDSDi output of the EX1 and EX3 is 10 bit.

I have no idea what the HDMI output is. However the EX's are a little on the noisy side so any advantage that recording in 10 bit may give is likely to be lost in all the noise. It will be the image noise that will govern how far you can grade, not the bit depth, unless you have a very quiet camera like the F3.

[Olakunle Olanrewaju]

I read from a link in dvx.com that output from hdmi port of ex1r is 10bit, but i want some independent confirmation from probably sony or the likes.Maybe Alister Chapman can help in that direction.

[Alister Chapman]

I wouldn't get too hung up on whether it's 10 bit or 8 bit because it really isn't going to make any difference whichever it is given the EX1R's noise figure. You just won't gain any advantage recording 10 bit over 8 bit with a noisy camera. You will get banding from the noise quantisation before you get banding from the lack of bit depth.

Besides which IMHO the extra noise that using ProRes with the Ninja will add isn't going to help the situation along. You'll certainly get better recordings using a Ninja than the internal Mpeg 2 at 35Mb/s, in particular if you shoot interlace but don't expect to see a big difference in the raw footage because you won't. It's only if you do heavy grading that the differences will show up and then you will be noise limited, not bit depth limited. In my experience ProRes adds noise during the encode stage.

[Cliff Totten]

Hmmm,... well? I never thought of the EX1R as being a high noise camera. (Although I certainly know that sensor technology has come pretty far in the past 2 years)

I'll be running under tons of lighting in front of a green screen, prolly with negative gain. (It's a rock video)

Alister, do you really think that 10 recording is a waist with the EX1R under strong lighting conditions? (if the EX1R is truly "10 bit")

I think that recent developments in image processing is forcing us all to reevaluate what we call "high gain" these days. 9db on an EX1r is "high". That same 9db on an FS100 is nothing! (I'm shocked at how good 15db looks on that thing)

Everything is relative I guess.

Thanks Alister.

P.S. Generally speaking, do you give the Atomos Ninja your "thumbs up"?

[Alister Chapman]

Don't get me wrong. I don't think the EX1 at 0db is excessively noisy, it's just compared to many other cameras it is noisy. Taking Sony's own noise figures (If I remember correctly) the EX1 is 54db, the PDW-700/F800 and PMW-350 are 59db. 6db is a halving of the amount of noise, so these camera have almost half the noise of the EX's and you can clearly see it in the images, in particular if you grade heavily where the difference is night and day.

The issue with these cameras is that the noise is distributed evenly across the entire image. It's in the highlights as well as the low key areas and this makes grading problematic. Unless you like increasing the amount of data storage you need, there is little to be gained recording an EX using 10 bit over 8 bit. This has been discussed many, many times in the external recorders forums. The noise modulation is larger than the 8 bit sample size, so all that 10 bit does is waste data on more accurate noise reproduction.

For the moment I'm on the wall with the Ninja, in all the unit's I have played with so far the firmware has been too buggy for my liking. I've yet to see one play anything back correctly and if you can't review your material on location then how can you be sure everything is working correctly? Perhaps later firmware builds will resolve (or have resolved) this, it certainly shows promise, but it's not for me at the moment.

[Olakunle Olanrewaju]

Anyway for whatever its worth I really want to know what the bit depth of HDMI port in ex1/ex3 cameras is so that
at least that Question could be put to rest Ninja or no Ninja.

[Barry J. Anwender]

Originally Posted by Alister Chapman

Don't get me wrong. I don't think the EX1 at 0db is excessively noisy, it's just compared to many other cameras it is noisy. Taking Sony's own noise figures (If I remember correctly) the EX1 is 54db, the PDW-700/F800 and PMW-350 are 59db. 6db is a halving of the amount of noise, so these camera have almost half the noise of the EX's and you can clearly see it in the images, in particular if you grade heavily where the difference is night and day.

Alister, this raises a good point based upon the findings in the BBC WHP034-ADD64_rev3_Sony_PMW500 report. The PMW500 has the same CCD's as the 700/800. Alan Roberts, the author measures a sn of 54db with "noise suppression" off and a sn of 59db with "noise suppression" on. Recently, B&H is now also quoting these two sn figures. Alan goes on to say, "The price to be paid for this reduction of noise levels is a little softening of the picture." and he concludes, "Noise performance is ambiguous."

I am considering the purchase of PMW500, but the BBC report muddies the water. I want a sharp picture, thus if 54db is a good as it can get, then I will stay with my EX3.

[Alister Chapman]

The PDW-700/PMW-500 is much quieter than the EX cameras, NR on or NR off. In most cases the image softening from the NR is barely noticeable. According to Alan Roberts the EX1's noise figure is actually 44db, so big difference if you go by Alan's numbers.

However not all noise is created equal. Some noise is a lot more objectionable/noticeable than other noise. It depends on how fine it is and how "busy" it is.

[Barry J. Anwender]

Yes, thanks for that qualifier Alister. Allan Roberts EX1/3 report does point out a sn of 44db and some insight to provide an answer to the original question of this thread:

"Clearly, the noise levels change well in accord
with the slope of the gamma curve, there being
about 10dB between black and white. This
indicates that the camera front end is the prime
source of noise, and that the adcs are probably 12-
or 14-bit. Since the slope of the gamma curve is
unity when the signal level is about 50%, it is also
clear that the noise level is only about -44dB rather
than the claimed -54dB in the specification."

Sony has previously gone on record to clearly state that the EX1/3 outputs are 10-bits. Alister has put some real world observations into the effectiveness of these 10-bits.

[Cliff Totten]

Considering this noise-floor talk all is indeed true, is there any advantage to running at negative gain?

Or, will that not change the inherent noise characteristics at all from 0db?

I would expect that if you ran at -3db, you could expect a 3db reduction in noise? Or....hmmm,...does that increase the overall signal to noise ratio by 3db? (or is that just wishful thinking?)

CT

[David Heath]

Negative gain may have a positive benefit on noise (though it's doubtful whether it will substantially affect the 8-10 bit debate) but is quite likely to have an adverse effect on highlight performance.

In simplest terms, the max output from the sensor should be far more than necessary to give 100% exposure - the difference can be processed to retain detail in the highlights. Use negative gain, and a higher chip output is required for basic exposure, so less headroom available to retain detail in the highlights.

[Steve Kalle]

According to Dan Keaton from Convergent Design, all HDMI outputs on today's cameras are 8bit.

Some have posited that the HDMI outputs are only 8bit because all HDMI connections on TVs are only 8bit. So, these 8bit HDMI ports are mass produced and cheap whereas a 10bit HDMI port would be expensive and have low demand.

[Alister Chapman]

I think the notion that the HDMI output is only 8 bit due to the use of mass market 8 bit components is a valid one. When the EX1R was developed there weren't any external 10 bit HDMI recorders and 10 bit HDMI monitors were zero, so there would have been no foreseeable need to use more expensive or specially developed 10 bit devices.

I don't think many people realise that the vast majority of LCD TV panels are only 6 bit. They have a 8 bit input to give a reasonable overhead for processing and gamma correction before display on a 6 bit panel. 8 bit panels are appearing now, but are still not common, 10 bit even rarer. Many of the banding issues we see in our computer based edit suites or on TV are due to these 6 bit panels.

Negative gain reduces dynamic range as it will artificially clip your low key parts of the image. The brightness range is restricted by the cameras DSP, it only has so many bits to sample the sensors output, from dark to light. If you reduce the gain the highlight handling doesn't change (DSP bit depth limited) but your entire image get shifted down and as a result you will clip off some of your shadow and dark information, so your overall dynamic range is also reduced. 3db is half a stop (6db = 1 stop) so -3db gain reduces the dynamic range by half a stop.

Interestingly the PMW-F3 has an excess of dynamic range for the normal gammas and cinegammas and the processing appears to take advantage of this to keep the images very clean. When you shoot with the standard gammas and cinegammas on the F3 the cameras base ISO (sensitivity) is 400 asa at 25p. This gives a very clean, ultra low noise image with 11.5 stops of dynamic range. When you switch the camera to S-Log, which gives a greater dynamic range (approx 13 stops by my estimation) the base ISO increases to 800 asa. Looking at some of my S-Log test footage a clear increase in under exposure latitude can be seen when you use S-Log. I suspect that the "0db" point in the F3 is actually 800 asa as used by S-Log, where maximising dynamic range and using the full sensor range is the priority. Meanwhile with standard gammas, which are limited to 11.5 stops anyway, you can reduce the gain by 6db (1 stop) sacrificing one stop of underexposure but still have the full 11.5 stops but with 6db less noise.