vDSLR @ ISO 6400 vs. 18db gain on Consumer and Prosumer Camcorders

[Ripple]

Thanks to those of you who recently helped here with my general questions about Canon vDSLR’s.

1. I am wondering how the t2i & 7D at ISO 6400 compares the consumer Canon, such as the HV20 series at 18db of gain (I have an HF100, which I believe uses the same sensor).

2. I am also wondering how the t2i & 7D at ISO 6400 compares with the Sony and Panasonic’s prosumer camcorders at 18db of gain (i.e. the PMWEX1R, PMWEX3, HMC150, HVX200a, and HPX100).

Edit: The following may or may not be useful...

The reason I am asking is that low light performance of the video of Canon vDSLR’s may be the tipping point for me when choosing between Nikon and Canon for stills (I do not think Nikon is better for stills in absolute terms, but it is likely better for me personally).

I talked to a professional photographer who said the video from his 5D Mk. II looked less noisy than his HMC150 with just 6db of gain, but I never saw the footage.

I have read that the sensor of the Canon HV20 is rated at ISO 100. With 18 db of gain (three stops), I think this would be like ISO 800. So if the 7D or t2i were similar in noise at ISO 6400, this would give me an extra 3 stops (at equivalent apertures).

IIRC, the sensors of Sony’s PMWEX1R and PMWEX3 are rated at ISO 400, as are and the sensors of both the Panasonic HMC150, HVX200a, and HPX100 (please correct me if I am wrong here). At 18 db of gain, this would be ISO 3200. The reason I am asking about these prosumer cameras is that I figure people here may have more experience with them than the consumer Canons, and I may be able to subtract 2 stops to get a hint of how noise would look on a consumer Canon (native ISO of 100 vs. 400).

Also, I understand that not all cameras handle 18db of gain the same. If there are substantial differences in how gain is handled between, consumer Canons and various prosumer Sony and Panasonic camcorders, it would be interesting to know.

BTW, I realize the t2i and 7D have different sensors, but both look to me to be within a half stop of one another at higher ISO’s (more green noise in the t2i though). At any rate I will be waiting to see what the successor to the 50D looks like this fall before investing $2500 into a DSLR system (body and lenses).

 

[Barry_Green]

Whether you call it "gain" or "ISO", what you're really talking about at the end of the day is electronic image amplification. And they are translateable, you can translate a certain gain figure into becoming an equivalent ISO value on the video camera. Each 6dB of gain amounts to doubling the ISO.

So your first question: 6400 ISO vs. +18dB on an HV20 -- the DSLR will win that sensitivity battle in a landslide, because the HV20 starts with such a low ISO in the first place. The HV20 is 80 ISO. And with 18dB, it becomes an effective 640 ISO (80 +6dB = 160, +6dB = 320, +6dB = 640). The difference between 640 and 6400 means that, yeah, for equivalent sensitivity the HV20 would need 10x as much light. As for grain -- you'd have to test that. The DSLR isn't too clean at 6400 ISO, but the HV20 isn't very clean at 6dB, much less 18dB. And, the HV20 can go a lot higher than +18dB; if I remember properly it can go to at least +27dB. Which would result in a total max "iso" of about 1920. Still not a third of what the DSLR could do.

The EX1/EX3 is 400 ISO when in 1080p mode, 500 ISO when in 720p mode, and 800 ISO when in 1080i mode. So for a direct sensitivity comparison, 1080i at +18dB = an equivalent of 6400 ISO.

The HPX170/HMC150/HVX200A are 500 ISO. So at +18dB they're equivalent to 4000 ISO.

You might find this older thread helpful; this is where I went and put an HPX170, GH1, and 7D up against each other testing noise/grain at high ISO/gain levels.
http://www.dvxuser.com/V6/showthread.php?t=190197

 

[sublimesnfu]

Thanks for the info :-D

 

[Ripple]

So your first question: 6400 ISO vs. +18dB on an HV20...

....The HPX170/HMC150/HVX200A are 500 ISO. So at +18dB they're equivalent to 4000 ISO.

You might find this older thread helpful; this is where I went and put an HPX170, GH1, and 7D up against each other testing noise/grain at high ISO/gain levels.
http://www.dvxuser.com/V6/showthread.php?t=190197

I looked at your test. I will read the rest of that thread when I have time.

If I am interpreting you correctly, the 7D at ISO 4000 with in camera noise reduction has a similar amount of noise as the HPX170 without in camera noise reduction. Is this correct?

Edit: the HMC150, HVX200a, and HPX100 all use the same sensors, yes? [end edit].

Gain/high ISO noise aside, IIRC, it has been said or implied that the HMC150's implementation of AVCHD looks significantly better than Canon's vDSLR codec (even though both are H.264 and the Canon uses a higher bits/s). Would the people here agree with this?

(The reason I ask is that a used HMC150 is the probably only prosumer camera that may come into my budget in the next year or two, after the Scarlet 2/3" and the next gen of Japanese camcorders drives down used prices).

 

[Barry_Green]

If I am interpreting you correctly, the 7D at ISO 4000 with in camera noise reduction has a similar amount of noise as the HPX170 without in camera noise reduction. Is this correct?

The images I got are posted, so you can decide for yourself. I think they're all grainy and noisy at that speed.

Edit: the HMC150, HVX200a, and HPX100 all use the same sensors, yes?

Yes, those three all use the same sensors. But the HVX200A is tuned slightly differently, resulting in a minorly softer image at equivalent settings; you can change the settings to equate the pictures though.

it has been said or implied that the HMC150's implementation of AVCHD looks significantly better than Canon's vDSLR codec (even though both are H.264 and the Canon uses a higher bits/s). Would the people here agree with this?

Doesn't matter whether the people agree or not, all that really matters is what the facts are. And, talking strictly as a codec, yes the HMC150's AVCHD is substantially better and more robust than the Canon's codec. The Canon uses a limited subset of h.264 that doesn't provide for B-frames, and it has to throw 2.5x as much bandwidth at the issue in order to get reasonably robust footage. The AVCCAM implementation of AVCHD uses the full toolset including B-frames, and it gets more robust resilient artifact-free imaging out of only 21 megabits.

Seriously though -- you really would have to compare the two of them yourself to see what you prefer. Whether one has more pixels or less grain or a better codec or whatever is fine for paper comparisons, but if you just plain like the look of one better than the other, then the specifications might not count for so much. Or if you need the functionality and features that one offers, that the other doesn't. Or if you need the price tag that one has, and the other doesn't.

 

[Ripple]

The images I got are posted...

Yes, those three all use the same sensors...

...The Canon uses a limited subset of h.264 that doesn't provide for B-frames, and it has to throw 2.5x as much bandwidth at the issue in order to get reasonably robust footage. The AVCCAM implementation of AVCHD uses the full toolset including B-frames, and it gets more robust resilient artifact-free imaging out of only 21 megabits...

Thanks for your patient and thorough explanations. Also, interesting about the specifics of Canon's vDSLR codec vs. Panny's in implementation of the AVCHD codec.