Ian-T
Veteran
I was waiting for someone to use that line...lol
Low Light Gh1 Vs Dvx100a
- Thread starter Slice
- Start date
No, it is not just ISO and f-stops.
Not at all actually....
Noise is about light gathering, nothing else (well, certain companies treat noise better than others, but in the end of the day, light gathering is what counts).
Light gathering depends on three parameters:
1) The f-stop
2) the sensor area
3) the exposure time
In videography the exposure time of each image is basically the same, so that equals out between different cameras.
The f-stop is what you get if you divide the focal length with the diameter of the entrance pupille of your lens. The faster the lens the more light hits the sensor. The Gh1 kit lens is indeed rather slow, but nothing prevents you from using a f1.4.
The most important factor is the sensor area. A bigger sensor will gather more light and the more light you have, the less noise you have as well.
This is the real reason for why a 5d2 shines noisewise..
The sensor is so much bigger than in traditional camcorders so its like a sumo fighting mr bean.
If a camera has a sensor that is twice as big as another, it will gather the same amount of light with a lens that is one stop slower. That is the simple math.
So what the hell is the ISO about?
The iso is measure of how much you amplify the signal (=the light that you have gathered) to create your final image. If you crank up the iso you increase the amplification of the signal. The problem is that you amplify everything, also the noise which will become visible.
A camera/camcorder with a big sensor can accept a higher ISO-level than a camera with a smaller sensor simply because there is more light gathered (stronger signal) and the signal to noise ratio is higher.
It's exactly like with a radio: If you have a good antenna (=large sensor) you catch the station well. If you have a small antenna (=small sensor) you have to raise the volume, but by doing so you also raise the background noise and it doesn't sound good.
The gh1 is almost certainly much cleaner at ISO1600 than a traditional camcorder at ISO 640 because of the bigger sensor.
The comparison of ISO levels between different cameras is only interesting if you also consider the noise levels.
Yes, I generally agree, but instead of sensor size, I prefer to focus on the effective light gathering capacity. Eg. Bayer sensors (CCD & CMOS alike) waste 2/3 of the light in the colour filters. 3-chip designs have a much lower loss rate, and concerning the 5Dii, allegedly there is an aditional 2/3 loss of signal due to the crippled readout method.
IOW: sensors size, f-stop, sensor efficiency, readout method are equally important.
These are the 'objective' parameters that can be compared easily, the 'subjective' ones are the amplifiers, A/D converters, the noise reduction method etc. also affect the output greatly.
Actually, contrary to common belief, the pixel count is probably the least important aspect. With higher pixel count, the noise on pixel level goes up, but resampled to common denominator the noise stays remarkably constant. So, as long as the higher pixel count is properly resampled (not like the 5Dii), it's fine.
John Caballero
Veteran
Blah, blah, blah, blah.
Has it been tested? The fact that the 5Dii doesn't completely smoke the GH1 for high iso and only minor aliasing issues reported suggest that GH1 is more efficient with the smaller sensor. But if you have a source I'd love to see it.
i have no evidence but it has been said before, the gh1 has far better low light than current video cameras (EX1)
http://www.dvxuser.com/V6/showpost.php?p=1632911&postcount=1381
http://www.dvxuser.com/V6/showpost.php?p=1632911&postcount=1381
jenningsp
Well-known member
the GH1 is excellent in lowlight. it's better than and ex1, it's better than a DVX but only if you use a super fast lens.
if you use the stock lens, it's gonna be worse in lowlight than the DVX - and by this i don't mean noise/grain wize. i mean just being able to see something. i'm sure a pitch black image is super clean and has very little noise
also with the stock lens you're going to get about the same DOF as a DVX as well. so what's the Fing point......
stock lens is just slow. deal with it. wish they sold it without it.
according to amazon japan the cam by it's self is only worth $538
if you use the stock lens, it's gonna be worse in lowlight than the DVX - and by this i don't mean noise/grain wize. i mean just being able to see something. i'm sure a pitch black image is super clean and has very little noise
also with the stock lens you're going to get about the same DOF as a DVX as well. so what's the Fing point......
stock lens is just slow. deal with it. wish they sold it without it.
according to amazon japan the cam by it's self is only worth $538
Last edited:
I can't point you to an official source, but this as been discussed earlier on this board, the GH1 use Point resampling, not Bilinear/Bicubic. However, considering the max sensor resolution is 4352 (width) whereas the 5Dii is 5616, resampling to 1920 produce less artifacts. The evidence of this is that GH1 produce some artifacts when filming horizontal/vertical lines, but fewer than 5Dii.
..So the GH1 has better low light capabilities than DVX. What you're saying here is that DVX has more gain and this can see better in low light (are you sure about that, given the low light footage with the stock lens we have seen so far ?), but you can boost the gain in post on GH1 footage, so I think it still win even with the stock lens
ydgmdlu
Veteran
You did get right that the main factor in low-light performance is light-gathering ability, but can you please explain why greater sensor area leads to greater light-gathering ability? Do you not understand what an imaging sensor actually is? It's an array of photosites. The sensor itself does not capture light, per se; it's the photosites on the sensor that do it.
Let's do a thought experiment. Imagine a 2.1 MP wide-aspect sensor. (This would be a 1080p HD sensor, FYI.) Say that it's a 1/3" chip, like the majority of prosumer camcorders. Now take those same 2.1 million photosites and spread them evenly across an area 36mm wide by 20.25mm high. (This is the effective imaging area of the 5D Mk II's video mode.) If the same number of photons are hitting both sensors, then how should the bigger sensor perform compared to the smaller one?
If you insist that the bigger sensor in this scenario somehow gathers more light, then you must explain how. Or to put it another way: How would the exact same photosites manage to gather more light simply by virtue of being spread across a much larger area?
If we grant that a bigger antenna picks up a stronger radio signal than a smaller one, which is a dubious and simplistic claim, then the reason why is because the entire body of the antenna is acting as a conductor for the signal. An antenna is really more analogous to a photosite. An imaging sensor is not a monolithic light-sensing device. What we know as the "chip" is made up of millions of tiny sensors. The output of the chip is essentially a mosaic of the read-outs of the individual photosites. This is why the size and performance of the photosites matters most, not the area of the sensor. The area of the sensor is almost meaningless, except to the extent that it allows the use of bigger lenses, which of course gather more light.
You're still going on about this? Yes. IF the same number of photons hits the big sensor then it isn't any better than the smaller one. The thing is that the same number of photons is NOT hitting the sensor. If you use the same lens (which is capable of covering both sensor sizes) the number of photons collected by the big sensor is much greater. The photons collected per area is the same, but the big sensor has more area. It's funny that you seem to think that area only matters on the pixel level, but not on the sensor level. You can't compare one big pixel to one smaller one. You have to compare equivalent areas. One big pixel vs. several smaller ones.
I posted this link in previous thread, but you just chose to ignore it. You keep demanding proof, but it's you who hasn't proven anything.
Here is the link once again: http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/index.html
Quotes: "[FONT=Helvetica Neue Bold, sans-serif]Nevertheless, what should be clear from the preceding analysis is that there is virtually no difference in photon collecting efficiency over a very wide range of pixel sizes, from 2 microns to over 8 microns."
"[/FONT][FONT=Helvetica Neue Bold, sans-serif]The above DSLR/digicam comparison outlines the extremes of what may be possible with current or near-term technology, if digicam pixel densities were used to populate full-frame sensors. The fact that a digicam's performance is in the same ballpark as the best DSLR's when referred to fixed spatial scale, suggests that the problems with noise in digicams is not due to their ever smaller pixels, but rather it is due to their continued use of small sensors."[/FONT]
The 5d2 has 21.1 Mp on 24x36=864mm2 sensor area. This means that 1080x1920 pixels take up 85mm2. The size of a 1/3" sensor is 4.8 x 6= 28.8 mm2
So, the effective light gathering surface area of the 5d2 is log(85/28.8)^2/log2 = 0.98 step bigger (where each step is a doubling).
The horrible thing is that we still live in camcorder stone age because the 5d2 could have used ALL PIXELS on the sensor to gather the light and then use pixel binning to output a 1080x1920 signal.
In this case the difference to a 1/3" sensor would be a whopping
log(864/28.8)^2/log2 = 4.9 step advantage..
The video function on todays DSLRs is tacked on, but the result is still better than on prosumer camcorders. It's sad that they didn't go all the way, but both canon and panasonic have a camcorder division to protect...
So, the effective light gathering surface area of the 5d2 is log(85/28.8)^2/log2 = 0.98 step bigger (where each step is a doubling).
The horrible thing is that we still live in camcorder stone age because the 5d2 could have used ALL PIXELS on the sensor to gather the light and then use pixel binning to output a 1080x1920 signal.
In this case the difference to a 1/3" sensor would be a whopping
log(864/28.8)^2/log2 = 4.9 step advantage..
The video function on todays DSLRs is tacked on, but the result is still better than on prosumer camcorders. It's sad that they didn't go all the way, but both canon and panasonic have a camcorder division to protect...
gastrolit
The 5d2 has 21.1 Mp on 24x36=864mm2 sensor area. This means that 1080x1920 pixels take up 85mm2. The size of a 1/3" sensor is 4.8 x 6= 28.8 mm2
5Dm2 takes 36*20~=720 mm2 in 16:9, then reads every third line from sensor and downscales it to 1080p, AFAIK. So, effective square is 720/3=240 mm2. Comparative with ~200mm2 for GH1 in 16:9 mode.
The 5d2 has 21.1 Mp on 24x36=864mm2 sensor area. This means that 1080x1920 pixels take up 85mm2. The size of a 1/3" sensor is 4.8 x 6= 28.8 mm2
5Dm2 takes 36*20~=720 mm2 in 16:9, then reads every third line from sensor and downscales it to 1080p, AFAIK. So, effective square is 720/3=240 mm2. Comparative with ~200mm2 for GH1 in 16:9 mode.
do you have a gh1?
Ha, I've been on here for years and seen this happen countless times. If it looks and smells like it, it's measurebating...
Still, a couple weeks away from having cameras in hands so I understand the temptation. I just hope people realize that it's pretty much a pointless exercise in debating until side by side tests can be done.
Still, a couple weeks away from having cameras in hands so I understand the temptation. I just hope people realize that it's pretty much a pointless exercise in debating until side by side tests can be done.
John Caballero
Veteran
Precious.
marlontorres
Well-known member
WHO CARES?! This is like debating the length of an artists pencil! It's not the pencil that creates the work its the artist! This debate just makes my head hurt!