No, never said that. The level of a single photosite noise is irrelevant for our discussion. I do not not know how to say that in yet another way.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
7d and who needs those specifications?
- Thread starter boriszulj
- Start date
Well, then I guess you don't need much. Just buy a $600 entry level DSLR with moderate specifications and be done with it. Maybe that's why camera manufacturers make various models with various levels of specifications and features! So people can choose the camera that best fits their needs! Have you have thought of that?
18 MP on APS-C - don't need and very skeptic about that (considering 40d vs. 50d)
HD video - I have great Sony Full HD camcoder for that, smaller lighter, designed to shoot video, unlike DSLR that is designed to make photos
19 AF points - I use central point 90% of time
Live view w Face-priority AF - never use Live view cause it is pointless without articulated LCD (I' have Olympus c-8080 and know how that feature is useful)
8 fps - don't need cause don't shoot sports
Dual DIGIC 4- as I said, don't need 8fps, also processor that enables that
New 24mm built-in flash - I always use external flash
New LCD - I'm not buying TV
New iFCL metering (exposure and colour) - we'll see about that is it better or worse
And at the end, Aero-dynamic body design is the key reason to buy this camera!
--
Boris
Degustibus non disputandum est!
ejmartin
Veteran Member
Show me any sensor (production or even in the laboratory) with 2µ pixels and ≤1 electron of read noise, that would exhibit the theoretical scaling you favor, of 8µ pixels with 4 electrons of read noise (5D, 1D2, D3), and I'll shut up.
--
emil
--
http://theory.uchicago.edu/~ejm/pix/20d/
Greg Pavlov
Senior Member
What did you mean, then, when you said the following?
"This has nothing to do with the ignorant religious belief that small photosites collect less light and therefore are noiser."
Or do you share that "ignorant religious belief?"
You forgot to cite my second sentence. I am so pi$$ed off at this fallacy, so I did not want to repeat it. OK, I will do it now. The dogma goes like this:
1. Smaller photosites gather less light, and are because of that are noisier.
2. Therefore, a sensor of the same size with more pixels is noiser.
OK, (1) is correct, and (2) is wrong.
Now, the second part of (2) (without the "therefore" part) may still be right or not. Read noise, heat, and all that. This does not make (1), (2) right.
Greg Pavlov
Senior Member
I agree. I probably won't buy one either, at least not until discounting kicks in. On the other hand, if Canon came out with an f5.6 500 or an IS version of a f5.6 400........
...have fully functional cameras already. Why should I buy new one. Just to have "the latest gear"???
--
Boris
Degustibus non disputandum est!
bronxbombers
Forum Pro
although I am not entirely sure, I believe he is not talking about huge differences either way, certainly for something like a 10MP vs 20MP cam. So the giant increase in detail you can pull at ISO100 and perhaps a trace better per image DR there probably outwieghs a 1/3 stop or whatever loss at ISO1600. Although I'm not sure if any of us knows exactly how large the difference would be.If it is "actually true" it does not seem logical - or constructive - to call it "ignorant religious belief."
What he said and demonstrated seems to be quite clear: he claims that larger photosites + perform better than an aggregation of small photosites + additional software manipulation in low light, while the opposite is true in good light. I am not saying that I agree or disagree with that, simply that that is my basic understanding of his presentations. He does not make any claims, as far as I could find, that all real-world sensors of identical size and reasonably similar technology collect and are able to process the same amount of light. His starting point is that they do. That's not the same thing.
Most folks who are really into this stuff would like to see further technological improvement and you can pretty well safely bet that it will occur. However, there is a good chance that such improvements will not change the relationship in relative utility, and ideal utility, among sensors with varied densities of photosites. In other words, what is likely to improve the performance of a densely packed sensor is likely to improve the performance of a less densely packed sensor. Then again, maybe not. In the meantime, it appears to me that there are a decent number of alternatives out there to let folks choose what they need/want (mostly want...
Of course, at some point in time, the pain of dealing with large files may become worse than the ever smaller returns. Whether the ideal point for the avg person was met at 15 already or whether 18 may do it or go jumping the shark right past it I'm not sure. My guess it that it is certainly pushing the limits of cost/benefit, and maybe a tad over.
bronxbombers
Forum Pro
18% gray starts to be bright enough that photon collection probably means more than read noise and 50D has gapless collectors.
bronxbombers
Forum Pro
hmm I thought the 5D had a fair amount of pattern banding
the 5D2 does and yet I thought the 5D had even more
the 40D/1dmk3/1dsm3 (and perhaps 500D from what I hear) seem to be the best in terms of that of any of the recent canon designs
ejmartin
Veteran Member
I've followed this theoretical argument; the question for me is whether other constraints get in the way of its applicability in actual devices. I don't know what the root cause was of the fact that Canon was for a long time unable to reduce read noise with pixel size; the fact remains, that they were not. CCD digicams with ~2µ pixels have read noises larger than Canon DSLR's; different readout technology, to be sure -- but the SX1, with CMOS sensor and ~2µ pixels, has even worse read noise judging by images.
Yes, and I have consistently made this point (see for example my article).
Should -- but doesn't seem to in practice.
If other design considerations are more important, then nice theoretical arguments about how read noise should scale are moot.
I believe that DSP got his data from an analysis of DxO data, which is a bit suspect for the D90. I'd rather substitute the D300 (same sensor, same pixel design), whose ISO 1600 read noise is over 8 electrons.
Also, don't forget that the D3x needs higher data rates to cope with more pixels per column, which will give a hit to its read noise.
I doubt that differences of .1 electron in read noise are statistically significant.
--
emil
--
http://theory.uchicago.edu/~ejm/pix/20d/
KWEnz
Forum Enthusiast
I don't see lens micro-focus adjustment on the list (or even mentioned in these posts).... here's hoping it's included. That would be ONE of the most important features that I would like. Guess we'll know in 24 hours.
1. 18MP APSC CMOS
2. Dual DIGIC 4
3. 8 FPS (94 FINE JPEG/ 15 RAW)
4. New eight-directional double cross central AF point, total 19 cross AF points
5. ISO 100-6400 (Exp 12800)
6. 100% 1.0X magni VF with intelligent display, horizon assist
7. New iFCL metering (exposure and colour)
8. 3′ 920K Version II LCD, 160 deg viewing
9. Full HD video (w external mic input n HDMI output)
10. Live view w Face-priority AF
11. New switch for Live View/Video mode
12. Aero-dynamic body design, metallic shutter sound, 150K shutter life
13. New 24mm built-in flash, supports EOS intelligent speedlite system
14. A.L.O
15. P.I.C
16. Picture style
17. EOS Integrated Dust Removal system
Cheers,
Ken
1. 18MP APSC CMOS
2. Dual DIGIC 4
3. 8 FPS (94 FINE JPEG/ 15 RAW)
4. New eight-directional double cross central AF point, total 19 cross AF points
5. ISO 100-6400 (Exp 12800)
6. 100% 1.0X magni VF with intelligent display, horizon assist
7. New iFCL metering (exposure and colour)
8. 3′ 920K Version II LCD, 160 deg viewing
9. Full HD video (w external mic input n HDMI output)
10. Live view w Face-priority AF
11. New switch for Live View/Video mode
12. Aero-dynamic body design, metallic shutter sound, 150K shutter life
13. New 24mm built-in flash, supports EOS intelligent speedlite system
14. A.L.O
15. P.I.C
16. Picture style
17. EOS Integrated Dust Removal system
Cheers,
Ken
I could be 18mp on 1.45 crop and therefore equal to 15mp on 1.6 crop...just hope Canon learned more regarding processing
I want video in my DSLR..1080 at 24p for film making...I and my brother are bother looking forward to this.
I use center point only 25% of time the other 75% I use other AF points usually regarding the rule of thirds. Very important when using F/1.4 lenses.
I shoot sports 5 FPS has been good but I don't need 10 FPS so 8 is in the middle...sounds OK to me. Also when trying to capture natural expressions in portraits, I prefer higher frame rates.
I use on-board flash, Canon ETTL flash and wireless flash using Elinchrom wireless triggers, I like them all for there specific uses. I would very much like the new 7D internal flash to also act as a flash controller to my 580's.
I need a proper LCD that works outdoors in broad daylight.
I bet it is better by far
I have no idea what that even means.
The 7D is not made for you, just like the 1D3 is not made for me.
Then be happy with a 5D Mark II and be done with it!
That's correct, only the 50D lovers "knows" what they are doing, no one else has any clue. especially those from the 'other brand", LOL......
fyngyrz
Senior Member
Sorry, I'm talking about the 5DmkII, IE the current 5D. My fault for not being specific. Frankly, when the 5DmkII came out, I erased the 5D from any consideration. The 5DmkII does very well at ISO 25600 for my purposes -- astro stacking.
The 50D cleans the 40D's clock at high ISO (12800), because the 40D simply can't reach those settings in the field; you have no idea what you're shooting until you get to a computer and can push the results with a 40D if you try for 12800 push. Can't really speak for the '1' models... something about APS-H designs bother me, so I don't even look at them.
The 500D's samples seem to be pretty much banding free. I don't care if there's a little. But the 50D has a lot , and it gets in the way, though it isn't fatal. Stacking cleans it up.
The 5DmkII is clean at 25600, and that's probably where I'm headed next. Presuming another Canon camera release this year doesn't reach or exceed that kind of light performance. We'll see.
--
Photos: http://www.flickr.com/photos/fyngyrz/
Blog: http://fyngyrz.wordpress.com/
Malcolm Practice
Leading Member
That's pretty close to the negative proof that people always end up asking for. I provided an example of same family sensors where the scaling was super-proportional to sensel spacing at high ISO, which certainly doesn't support your argument. We both know that down at 2µ for CMOS,it's very likely that we are about to hit process limits, and it is probably the case that with current CMOS lines that's about the limit of scaling for current lines. New lines with finer geometries might reduce that some more. In the real world, designers have many options to juggle. If we talk about absolute abstract relationships of sensel properties with size, the assumption must be that they use none of them. To use the car analogy people seem to like, it is generally true, and few would argue, that engine power scales with capacity. Nonetheless, it has been possible for some engine designs to produce more power with smaller capacity than others.
--
Mal
Tonybl
Forum Enthusiast
I know some people are not interested in the video, but I had a top of the range Sony HD solid state video camera which did not take good viodeo compared to the 5DMII . Only Its sound was better.
The noise of a pixel doesn't depend linearly on the area of a pixel, but on the size of a pixel. As such, the global noise for a sensor of a given size depends on the size of a pixel, not on the sensor area.
http://clarkvision.com/imagedetail/does.pixel.size.matter/index.html
-----------------------------------------------------------------------------------
Consider two sensors of a given 20 * 20 millimeters size, one (labeled A) with 20 * 20 pixels and one (labeled B) with 10 * 10 pixels. Sensor A has a pixel size of 1 mm (and an area of 1 square mm), sensor B has a pixel size of 2 mm (and an area of 4 square mm).
Let's consider that 1 square mm from a sensor collects 100 photons.
Since a pixel of sensor A has a surface of 1 square mm, it collects 100 photons and has a noise of 10; an entire photo taken with it has a noise of 10 * 20 * 20 = 4'000.
Since a pixel of sensor B has a surface of 4 square mm, it collects 400 photons and has a noise of 20; an entire photo taken with it has a noise of 20 * 10 * 10 = 2'000.
As you can see, sensor A produce noisier photos than sensor B.
The relevancy of this matter remains for you to decide. If there is enough light to properly expose your subjects, more pixels may matter more to you because they provide more detail, despite the fact that the photo is noisier. However, in low light, less pixels and a lower global noise yields better photo quality.
http://clarkvision.com/imagedetail/does.pixel.size.matter/index.html
-----------------------------------------------------------------------------------
Consider two sensors of a given 20 * 20 millimeters size, one (labeled A) with 20 * 20 pixels and one (labeled B) with 10 * 10 pixels. Sensor A has a pixel size of 1 mm (and an area of 1 square mm), sensor B has a pixel size of 2 mm (and an area of 4 square mm).
Let's consider that 1 square mm from a sensor collects 100 photons.
Since a pixel of sensor A has a surface of 1 square mm, it collects 100 photons and has a noise of 10; an entire photo taken with it has a noise of 10 * 20 * 20 = 4'000.
Since a pixel of sensor B has a surface of 4 square mm, it collects 400 photons and has a noise of 20; an entire photo taken with it has a noise of 20 * 10 * 10 = 2'000.
As you can see, sensor A produce noisier photos than sensor B.
The relevancy of this matter remains for you to decide. If there is enough light to properly expose your subjects, more pixels may matter more to you because they provide more detail, despite the fact that the photo is noisier. However, in low light, less pixels and a lower global noise yields better photo quality.
antoineb
Veteran Member
And be announced on september 9th?
Also I believe many sources had indicated Canon as the manufacturer of the sensor for the Leica M9.
Doesn't this reinforce the possibility that the Canon 7D will indeed be full-frame, which would certainly fit with its boday that looks like it has the same width as the 5D?
Sure, the Chinese text says "APS-C", and "8fps". But if "APS-C" proves to be wrong, then maybe the 8fps is also wrong (it wouldn't be very fitting for a model below the 5D, to have not just better ergonomics but also better continuous shooting)?
One more day to wait.
Also I believe many sources had indicated Canon as the manufacturer of the sensor for the Leica M9.
Doesn't this reinforce the possibility that the Canon 7D will indeed be full-frame, which would certainly fit with its boday that looks like it has the same width as the 5D?
Sure, the Chinese text says "APS-C", and "8fps". But if "APS-C" proves to be wrong, then maybe the 8fps is also wrong (it wouldn't be very fitting for a model below the 5D, to have not just better ergonomics but also better continuous shooting)?
One more day to wait.
Similar threads
Keyboard shortcuts
- f
- Forum