Low Light Gh1 Vs Dvx100a

jenningsp said:
it's asinine to pass on information believed to be fact from a very reputable source?
Click to expand...

Why are you trying to put this on Hunter now? In your quote referred to by John Caballero, divide, Kholi and finally myself, you don't credit anyone as your source. It clearly appears to be your own statement.

In a different Cinematography related forum, the founder constantly reminds it's members not to speak beyond their experience; and when posting theory/conjecture to label it as such. Not only is that good advice for any on-line forum, but life in general. I think that those who ignore such warning do so at their own peril, and risk doing themselves unrepairable disservice.

I was actually just told a story last week, that can perhaps serve as a cautionary tail. A person of some on-line/website notoriety a year or so back has gone (essentially) dark these days because their professional reputation is entirely in the gutter, with no signs of improvement. The reason- they so repeatedly ignored the basic concept of not speaking beyond their experience, and passing opinion and theory (sometimes not even their own opinions and theories) as fact.

Amazingly, at one point this person managed to secure a job in the technical side of the business, and after a very short period of time- was fired because it became obvious to all that the job was completely beyond their working knowledge. I also understand that they overstated their qualification greatly during the interview process, which made the firing that much easier.


stephenvv said:
11? No way. I had 3 DVXs over the years and they are a noisy camera in cinegamma modes where you get the increased latitude. You could eek out 7 if you did not mind noise.
Click to expand...

If you'd like a seconding, my experience echos your own. I recall charting (a generous) 7.5 stops from the 100A before the picture became NG; and 5 of those stops were down.

I.

Illya Friedman
Hot Rod Cameras
 
Illya,

i really hoped this would die... but you keep instigating a response.

i see forums as a conversation, not a scientific journal. do you site your sources after every line you speak in real life? I was just passing on what i believed to be an established fact. After being corrected by kholi, I posted a quote and a reference, so your 'asinine' remark came across as a pointless dig.

this is a gh1 vs DVX100 thread. i own a DVX100. i have have done my own DR tests with a waveform. i know exactly what i'm talking about in reference to the DVX. and i'm comparing that information to the only DR test of the GH1 in existence today. Yes it's conjecture and theory. but that's what this thread is...

someone needs to do another DR test on the GH1 if Hunter is wrong. Illya, if you send me your GH1, i'd be more than happy to test it... what's that? you don't own one? :)

peace
 
Last edited:
SLoNiCK said:
Barry_Green

You can wire same color sensels via gap transistors. Nearly 13 additional transistors needed for 4x4 block of GRBG bayer sensor. Unlock transistors - and get unified charge for every color. BTW, it will give huge readout speed boost cause it's nescessary to read 3 charges vs. 16 in usual case. May be worth for high frame rate.
Click to expand...
Well, yeah, but -- that's hardly binning, is it? That's no more than row-skipping.

Binning is combining adjacent pixels into a "super-pixel". If you go skipping the adjacent pixels to link only the same-color pixels together, you haven't really created a bigger bin nor gained the benefit of a super-pixel. The idea behind a super-pixel, as I understand it, is to combine all charges hitting in one area into one pixel. But with gap transistors you'd be linking disparate pixels together, but not increasing the surface area of the one super-pixel you'd be (assumedly) trying to create.

So, I still say that pixel-binning, to gain the advantage of sensitivity and lower noise at the expense of overall resolution, is not a practical technique on a single-sensor Bayer camera.

Unless I missed your point?
 
My experience says the DVX delivers about 7 stops to the final DV tape.

Juan Pertierra said he could eke some more out of the sensors with his Andromeda, but it seemed like there was frequently an odd color cast to the extended range, which is most likely due to one of the sensors clipping while the others were still delivering a charge. For example, the highlights would go yellow-ish, an indication that the blue was already blowing out and the red and green were then delivering a bit more out of proportion.
 
jenningsp said:
i really hoped this would die... but you keep instigating a response.
Click to expand...

2-2-tango eh?

jenningsp said:
do you site your sources after every line you speak in real life?
Click to expand...

Well... If I didn't know something, I would say "I don't know", and not attempt to BS my way through it.

Are you really so peeved about being called on it? My bigger point was cautioning ANYONE who makes a habit of speaking first and thinking later. Is there really room in ones mouth for both an elbow and foot?

jenningsp said:
so your 'asinine' remark came across as a pointless dig.
Click to expand...

Really? I always think the term asinine being incredibly pointed, if you re-read what I wrote, you'll see I never pointed squarely at you.

jenningsp said:
i know exactly what i'm talking about in reference to the DVX.
Click to expand...

So far that's not my impression. Please demonstrate your methodology for deriving a usable dynamic range of 11-stops straight from the DVX100, and I will immediately declare you master.

jenningsp said:
Illya, if you send me your GH1, i'd be more than happy to test it... what's that? you don't own one? :)

peace
Click to expand...

Ok you've got a deal. Paypal me $1700 to cover shipping. Hope you can read Japanese.

Ja matta,

I.

Illya Friedman
Hot Rod Cameras
 
Last edited:
Illya, he didn't say that he gets 11 usable stops straight from the DVX. He said the sensors are capable of that much range, which is most likely a reference back to Juan Pertierra and the Andromeda. That would seem optimistic, however, as Juan's claims for the Andromeda were that it got as much as 9.5 stops, which Juan says represents a gain of about 2.5 stops over the original DVX (which would put it back at 7).

Let's please step back a moment and let things defuse, okay?
 
Last edited:
hahahahahahahhaha i love dvxuser.

i said the dvx has 7-8 usable stops and you can eek out a total of 11 from the sensor.

the extra stops aren't usable... but a technicality

but i can show you those 11 if you want. i'll be right back....
 
jenningsp said:
hahahahahahahhaha i love dvxuser.

i said the dvx has 7-8 usable stops and you can eek out a total of 11 from the sensor.

the extra stops aren't usable... but a technicality

but i can show you those 11 if you want. i'll be right back....
Click to expand...

Well it's my mistake for reading too fast. Until you clairifed, I thought you were trying to say that the DVX100 had 11 stops of dynamic range, which wouldn't surprise me after your other claim.

I.

Illya Friedman
Hot Rod Cameras
 
Last edited:
I love DR warfare. It's mysterious and contentious!

Let's lay bets on it...

I'll optimistically lay my bet on 8. Spin the debayer!
 
ok so here it is...

depending on your monitor you should be able to see the difference between grey and white in the first pic. and the difference between grey and black in the last.



dvxat28.jpg


dvxat8.jpg


dvxat11w64.jpg


one thing to note is that the ND filters are actually a little stronger than they're supposed to be. so this makes the DVX even more awesome...

f2.8 to f11 w/ 1/64 shows 11 stops of DR

here's the graph.....

f2.8 = JUST below white
f4
f5.6
f8
f4 + 1/8
f5.6 + 1/8
f8 + 1/8
f4 + 1/64
f5.6 + 1/64
f8 + 1/64
f11 + 1/64 (can still differentiate grey from black)
f16 = JUST above black


am i master now? :)
 
Last edited:
jenningsp said:
am i master now? :)
Click to expand...


I almost don't know where to begin. First things first.... I've counted to 10. Really.
.
.
.

Since these images are what you are presenting as proof of dynamic range, why don't you give us a little more technical detail about how exactly you went about this test, and exactly what conclusions it is that you're drawing.

Actually before you do that, why don't you take a look at this:

http://www.cinematography.net/compare 2007/index.htm

Now consider your test.


I.

Illya Friedman
Hot Rod Cameras
 
Last edited:
i miss counted when doing the test, so i only posted a grab of the 10 stop point... there is more room below that. there's almost 12 stops in total. but 11 realistically :) i'll post the extra grab now..

here's F16 + 1/64 ND - you can hardly see a difference between the grey and black but there is still a slight difference... this is just masturbation though :)

dvxat16w64.jpg


BTW i've read the link you posted before.

how i went about it...

set up a black,grey and white card on a wall.

set the camera to normal gamma (retains the most highlight info) and set master ped to -3 (which sets true black at 0IRE)

lit the grey card so it was JUST not clipping at f2.8 (so 99IRE). then just proceeded to stop down and add ND's and marked each fstop increment.

my conclusions are that the DVX has 7-8 usable stops and 11 in total. if you had to, you could save some badly under exposed footage with the After Effects exposure tool and neat video noise reduction :)

yay!
 
Last edited:
You can't compare Apple and Orange... DVX100 is 720*480, GH1 is 1920*1080.
Even if they had the same DR, GH1 has exactly 6 times more pixels, so if you downres GH1 to DVX100, you average 6 pixels for one, thus greatly reducing noise and enhancing DR.
 
8 pages and no one came to the idea to do a quick test shot comparsion instead of theorizing?

Frank
 
Patrick- I'm a reasonable guy, and I thought about what you said, about how you think of this is a conversation... so tell me what the point of this exercise was exactly?

Was it that it's possible to differentiate totally unusable grey from totally unusable black by boosting levels in software?

I don't need to tell you there is a big difference between usable dynamic range (the dynamic range that matters) and unusable dynamic range. Which most rational people would consider the closely adjacent area outside of usable. However, what you've done is show us some brutally soft, de-noised differences from the muck, raised-up and corrected in software.

Somethings are more "unusable" than "unusable". And I think you've just posted it.

Look, I'm a long time fan of the DVX100, but what you've done doesn't help the cause of the DVX100 "being awesome".

I'll offer up an EXTREME example- I have a crappy digital camera right near my desk. The camera has maybe 6 stops of usable dynamic range, in a 1.6MP imager. I just shot a simple chart, SEVERLY underexposed, and I pulled it up with only a gamma adjustment in XNview. Obviously no de-noise.

demod.jpg


Just because you can do this does not mean it's practical (or IMHO fair to include it as part of that "unusable" dynamic range).

Don't be offended- to be clear, I am not mocking you. On the contrary, I'm trying to help.

I'm going to use my sarcastic voice for a second just to prove a point-

OMG, there's a difference between pure white (far left), grey (bottom half) and black (far right) - holy crap this "Melmann 5000 Pro" must be the most awesomest camera of all time!

No of course it's not. This image stopped being "unusable" a long time ago, but just because there are still few photons deep in the CCD wells doesn't mean it's fair to call this extreme example of a piss-poor latent image part of the sensor's dynamic range.

By the way, with an F900 and a DaVinci, it's possible to do this same thing to 19-or-so stops underexposed from key!! It's one of the examples we did at the Santa Fe Workshops, where I've been fortunate enough to attend a couple, and help instruct at another.

I.

Illya Friedman
Hot Rod Cameras
 
i just wanted you to declare me master :) i showed you 11 stops of DR from the DVX.... come on.... you know you want to...
 
Last edited:
You must log in or register to reply here.
Back
Top