PDA

View Full Version : D90 vs. 5D Low Light



roxics
10-26-2008, 09:37 AM
I would assume the 5D is better in low light then the D90 just because it has a full frame sensor on it. But I'm kind of curious how much better it is in low light. Looking to pick up a D90 and curious to know how well it performs under street lights at night and in other low light situations I have seen the 5D footage tackle without issue.

Matthew Bennett
10-26-2008, 10:21 AM
The 5d seems significantly better (ie low noise) in low light...

Joseph Stunzi
10-26-2008, 11:45 AM
Getting a quantitative answer is going to be difficult because the 5D hasn't been released yet. However, overall I think the 5D2 is a better investment with the full frame, the larger photo capabilities, and the professional grade aspect of it. I guess it's all a matter of opinion but for the first generation of these guys... I'd say go Canon!

roxics
10-26-2008, 12:27 PM
I can't afford to go with the 5D2. Just wanted to know how much lower down the chain the D90 was going to be.

Matthew Bennett
10-26-2008, 01:06 PM
I can't afford to go with the 5D2. Just wanted to know how much lower down the chain the D90 was going to be.

Hold a bake sale...

I think though, that when you see the nice pristine, set-bitrate encoding of the 5d, plus the low light, plus the general sharp look of it, all over the web, etc, and then you go back to your d90, and see the codec bottoming out like the apocalypse was coming, and the stair stepping, and the weak banded, overall softy look it has, you will start to get bummed and want more.

But of course, if you're only content with doing web videos at a maximum size of 640x360 pixels, then the d90 has you covered... unless of course you're shooting anything moving, then you'll be bummed again...

PerL
10-26-2008, 01:46 PM
I can't afford to go with the 5D2. Just wanted to know how much lower down the chain the D90 was going to be.

Nobody knows yet. but my guess is about a stop.

Karsvall
10-26-2008, 01:51 PM
yeah, but the footage from the 5D looks like video. And its not just the framrate.

roxics
10-26-2008, 02:34 PM
But of course, if you're only content with doing web videos at a maximum size of 640x360 pixels, then the d90 has you covered... unless of course you're shooting anything moving, then you'll be bummed again...

Maybe you are right and maybe you are being too hard on the camera, I don't know until I have one in my hands. But the footage I've seen from it so far looks great. But then I've got a steady hand and a fluid head tripod.

Jesus Christ could show up at my door right now with a 5D2 and tell me how great the camera is but I still wouldn't have the cash for him. Too many other things to pay for.

I'm looking at the D90 right now for hobby photography, perhaps pro photography if I can get the work and short films of my own at DVD resolution. I haven't had a DSLR since early summer when I sold my Canon 300D. Compared to that camera I'm sure I'll be impressed with any newer model.

I would expect the 5D2 at almost three times the D90's price to be superior, was just curious how much superior.

I remember not all that long ago before the DVX100 came out when you couldn't find a 24p SD camera for under $4000. Now we've got the D90 that shoots 24p HD with 35mm DOF in low light for $1000. I'm so geeked! It might have its limitations but filmmakers have always had limitations and still found ways to shoot great stories.

Thanks for all your info guys.

roxics
10-26-2008, 03:04 PM
Ok on second thought maybe I was just blind before with my optimism for the D90, but what are these horizontal lines I see jutting across the D90 videos on vimeo from time to time? They are like flashes of horizontal lines every so often. What is that? What is that called? Is that just vimeo doing that?

ydgmdlu
10-27-2008, 07:40 AM
The 5D Mark II's low-light performance is groundbreaking. The D90's is merely good. Camcorderinfo finally did an exhaustive evaluation (http://www.camcorderinfo.com/content/Nikon-D90-DSLR-Camera-Review-35626/Performance.htm#) of the D90, and they found that while its low-light performance is strong, it's disappointing given the sensor size. Basically, it outperforms $600-1000 consumer HD camcorders at moderately low light (~60 lux) but falls apart lower than that, underperforming the consumer camcorders starting at around 15 lux.

ydgmdlu
10-27-2008, 08:11 AM
yeah, but the footage from the 5D looks like video. And its not just the framrate.
No, it really is primarily the frame rate. It's not that the 5D footage looks like "video," it's that it doesn't look exactly like film. It's not the standard "video look." Compared to film and video of the past, the 5D's footage is much cleaner, and the colors and dynamic range are closer to life. By these objective criteria, the footage should be considered superior, but unfortunately, that's not how so many people here feel. So much of the "film look" is because of film's flaws, and to somehow declare that this look is categorically superior to any sort of "video look" is exactly like saying that vinyl's (technically flawed) sound is superior to pristine digital audio.

Sure, whatever look is desired depends on taste and intended effect, but what I've recently come to resent is the notion that the film look is preferable in general. Instead, the goal should be the best images, regardless of whether the source is film or video. Just like with vinyl, this attachment to the "film look" is not because it's associated with objectively superior output, but simply because everyone has grown up with it. The time has come to judge the quality of any given footage on its own terms, not how well it emulates the look of a particular medium.

Anyway, to get back to the original point, if you don't believe me that the "video-like" quality of the 5D's footage is mostly because of frame rate, then just freeze a frame of any of the posted footage so far and tell me in what way it glaringly fails to look like film. I've also seen film that looks like video (most recently, Appaloosa). So can we please set these pathetic complaints aside?

dUNIT
10-27-2008, 08:25 AM
The 5D Mark II's low-light performance is groundbreaking. The D90's is merely good. Camcorderinfo finally did an exhaustive evaluation (http://www.camcorderinfo.com/content/Nikon-D90-DSLR-Camera-Review-35626/Performance.htm#) of the D90, and they found that while its low-light performance is strong, it's disappointing given the sensor size. Basically, it outperforms $600-1000 consumer HD camcorders at moderately low light (~60 lux) but falls apart lower than that, underperforming the consumer camcorders starting at around 15 lux.

That's because they only used the stock lens.

PerL
10-27-2008, 02:15 PM
That's because they only used the stock lens.

Yes, i was wondering what the light levels of 15 lux corresponds to in terms of ISO/shutterspeed/aperture.
Compared to the kit lens speed (F 3.5 - 5.6) a F 1.4 lens on the D90 has an advantage of 2 2/3 to 4 stops.

bearing
10-29-2008, 10:54 PM
I remember people were saying the D90 is "skipping" when downsampling the image. This means it's using one "sensor element" per pixel.

The 5dmk2 may be "binning" which means it will add the signal from several "sensor elements" to form pixels when downsampling. If it does it means the light sensitivity is equal to a sensor with "sensor elements" of size equal to the the total area of the "binned" "sensor elements".

I remember reading a statement that the 5D mk2 have way better low light capability than the Red One. That may be true, if Canon is binning pixels when downsampling the sensor.

The theoretical properties for different sensors:


Camera Sensorsize Res. "Image element area" Videores. Effective pixel area (binning)
D90 23.6x15.8mm 12.2MP 30 (http://elektronikforumet.com/forum/viewtopic.php?t=29847)m 1280x720 339 (http://elektronikforumet.com/forum/viewtopic.php?t=29847)m
5D mk2 36x24 mm 21MP 41 (http://elektronikforumet.com/forum/viewtopic.php?t=29847)m 1920x1080 351 (http://elektronikforumet.com/forum/viewtopic.php?t=29847)m
Red One 24.4x13.7mm 12MP 29 (http://elektronikforumet.com/forum/viewtopic.php?t=29847)m 4520x2450 29 (http://elektronikforumet.com/forum/viewtopic.php?t=29847)m

So, if the Canon is binning it got 12 times better low light capability than both the Red and the D90 (if D90 is skipping and if the light gathering efficiency is the same). It's the equivalent of about 3.5 stops.

However, the resolution of the Canon is lower. If images from the Red would be downsampled to the same size they would be one stop darker.

Remember though, I'm making conclusions out of assumptions.