sritri wrote:
I got a 7D as a wedding anniversary gift a couple of minutes ago ( I normally shoot with a 1Ds-II
But noise is terrible in this am I doing something wrong ? 2.8, 1/13, ISO 400 on a tripod...70-200 IS USM
You've probably realized by now that this is not a good place and time for practical jokes. People are on the edge, they lost not only the common sense but the sense of humor as well
No jokes. Practical or otherwise. I just opened the gift box to find the camera and snapped a few. I know I need to RTFM first but having used a Ds-II for a while now, I thought ISO 400 should be easy
Well, just remember that ISO 400 on 1.6x crop cameras is about ISO 1000 on FF, as far as photon noise is concerned (read noise will likely favor the 7D by a substantial margin over the 1Ds2, but that only enters in shadows at these ISO settings). Don't expect too much.
ejmartin wrote:
Well, just remember that ISO 400 on 1.6x crop cameras is about ISO 1000 on FF, as far as photon noise is concerned (read noise will likely favor the 7D by a substantial margin over the 1Ds2, but that only enters in shadows at these ISO settings). Don't expect too much.
Emil, if you don't mind me attempting to elaborate on what you are saying (please correct as you see fit):
The noise here is mainly due to the noisy nature of light itself (i.e., "photon noise," or "shot noise"). The pixel pitch of the 7D is actually allowing us to better resolve the intrinsically noisy nature of physical reality when viewed at such a fine scale.
If an equal area of the scene were to be imaged on equal areas (e.g., same number of square millimeters) of a 7D sensor and a 1Ds2 sensor, the 7D image, when properly down-sampled to the resolution of the 1Ds2 sensor (i.e., same pixel pitch), should look no worse than the 1Ds2 sensor, and in fact, should look better.
So if I understand Emil & Paul correctly, it is more important to expose to the right "always" with the 7D than it was with the Ds-II ? As most of my shootings will be without flash on horribly lit stages, I guess the Ds-II should be more versatile ? I have not read a single line in manual yet, but there should be some way to disable Auto ISO in P/AV/TV modes, as I see that barring day-light, Auto ISO always picks ISO 3200 for a s/s of 100 or above.
sritri wrote:
So if I understand Emil & Paul correctly, it is more important to expose to the right "always" with the 7D than it was with the Ds-II ?
The more light you can get on the sensor without clipping the highlights, the better. Of course, you don't want to ruin your photo with Tv and Av values that prioritize "more light" over all other goals.
As most of my shootings will be without flash on horribly lit stages, I guess the Ds-II should be more versatile ?
Given the same f-stop, if you can fill your frame on your 1Ds2, it should be better than the 7D for this application (IQ-wise). If you are seriously reach-limited, the 7D is the choice. If in-between (need to crop the 1Ds2 image a bit), then it's hard to generalize.
The 1Ds2 has the advantage of the much larger sensor (about 2.5x in area), despite the older sensor technology. The chart at the bottom of this page of Emil's, while not including the 7D, gives a light-collection-efficiency ranking of cameras considering equal percentages of the frames (e.g., filling the 1Ds2 frame vs. filling the 50D frame).
An image uncropped from each sensor will have the FF camera collecting light over an area 1.6^2 ~ 2.5 times larger, so is collecting about 2.5 times as many photons. It has little to do with pixel pitch (and note that the total number of pixels is about the same for each camera, so we don't have to get into quibbles about whether the images should be resized etc.).
Now, given that you will be shooting at high ISO (how high?) the electronic read noise begins to play a more important role relative to photon noise. High ISO read noise in the 7D is about 2 electrons, or half of that of the 1Ds2. So about 40% of the photons, and 50% of the read noise, so S/N advantage of about a factor 1.25 in favor of the 1Ds2 is much closer than at low ISO, where photon noise dominates and the S/N advantage is more or less the crop factor, a factor 1.6 in favor of the 1Ds2. In-between amounts of light interpolate smoothly between these two values.
This is my first SLR with video features, therefore I have a total noob question. I am using a Mac and am having great difficulty transferring over the video from my camera to my computer via the standard usb connection. Does the EOS utility not support video? If not do I need to use a CF reader to transfer over the original movie file or do I need some other 3rd party software? Any help would be great as I am sure the answer is painfully obvious.
Is the 7D significantly better than that? I have seen Emil's chart at http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/D300_40D_tests/ but I'm not sure how it relates to QE. It looks like cameras overall vary by more than a factor of 2 from the 10D era to the most recent designs, but obviously QE cannot exceed 100%. And I thought QE anywhere near 100 was achieved only in exotic backside-illuminated designs (I guess not so exotic anymore, given some recent Sony consumer chips).
markoakley wrote:
This is my first SLR with video features, therefore I have a total noob question. I am using a Mac and am having great difficulty transferring over the video from my camera to my computer via the standard usb connection. Does the EOS utility not support video? If not do I need to use a CF reader to transfer over the original movie file or do I need some other 3rd party software? Any help would be great as I am sure the answer is painfully obvious.
Mark - I have a 5D2, and use a card reader to transfer everything. I use PhotoMechanic, but I would think that Image Capture (included with OS X, and much-improved in Snow Leopard) should be able to transfer everything for you. Not sure why the direct USB connection would be having a problem, but a good card reader (one of the newer UDMA units from SanDisk or Lexar, etc) may help with that problem, and should be substantially faster, as well.
jbeale wrote:
Sensor geek question: has anyone measured quantum efficiency for the 7D? According to the below chart (claimed accuracy only 20% or so) the 300D had a peak QE near 50% in the green. http://www.pbase.com/terrylovejoy/quantum_efficiency_of_the_300d
Is the 7D significantly better than that? I have seen Emil's chart at http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/D300_40D_tests/ but I'm not sure how it relates to QE. It looks like cameras overall vary by more than a factor of 2 from the 10D era to the most recent designs, but obviously QE cannot exceed 100%. And I thought QE anywhere near 100 was achieved only in exotic backside-illuminated designs (I guess not so exotic anymore, given some recent Sony consumer chips)....Show more →
I don't think the 7D is dramatically better than the 50D/5D2 in terms of QE; we'll have to wait for tests that are photometrically accurate (eg DxO's "ISO" normalization). But according to Christian Buil, DSLR's have a rather dismal QE once you slap on the IR cut filter and CFA:
Thank you for that very interesting and informative link! On one of the 40D graphs, I notice that removing the IR filter helps the red, of course, but actually decreased the QE in the blue, and also changed the response curve shape slightly: http://astrosurf.com/buil/50d/canon_40d_40dm.png
Maybe there is some AR-coating issue involved... I presume any fluorescent response in the filter wouldn't be a significant contributor... all very interesting.
jbeale wrote:
Thank you for that very interesting and informative link! On one of the 40D graphs, I notice that removing the IR filter helps the red, of course, but actually decreased the QE in the blue, and also changed the response curve shape slightly: http://astrosurf.com/buil/50d/canon_40d_40dm.png
Maybe there is some AR-coating issue involved... I presume any fluorescent response in the filter wouldn't be a significant contributor... all very interesting.
I'm a bit lost here.
Shouldn't the sensor be more sensitive to IR if the IR filter is removed? That's the whole point of removing the IR filter? There's zero efficiency at 700 nm.
Heechee - The respose at 700 is still filtered by the red CFA filter... But if you extend the chart, you will see a "backswing" of sensitivity in all channels out towards 800nm.
Qe depends a lot on how you define it. The most "fair" way to measure is to average a filtered [R,G,G,B] quadruplet over area. On average here you get to about 0.3 to 0.35 if youre measureing perfectly perpendicular D50 light. Gains here are very seldomly a "good" thing - most of the increase in Canon's later camera models stem from a dilution of the colour filter steepness - which gives lower colour resolution, a very different chroma noise structure and increased problems with metamerism. We've come as far as being able to get values as high as 0.8 (80%) when measuring a "normal", not BSI sensor, when measuring without colour filters - but with microlenses. BSI is mostly apllicable for small sensors, I don't see a reasonable cost-efficiency in performance-increase/price for yet another doubling of resolution in larger sensors.
Main things affecting total Qe are:
*Effective Fill Factor
*Silicate cell structure contaminations and physical layout (tunneling)
This is though not the only things that effect the effective Qe. Qe has a positive effect on noise on ALL levels of exposure at ALL ISO-amplification equivalents, signal to noise ratios increase all over the board. One problem you might run into is an upper limit in how big you can make the supporting "well", the charge capacitor that is meant to gather the electrons before converting from charge to voltage before A/D conversion. Nikon did run into this a while ago, and they start at ISO200 now. The FF Nikons are RN and amplification noise limited up to over ISO400 though, so it does not make that much of a difference, except if you really need long shuttertimes.
Effective Qe in a real situation always goes DOWN as the pixels get smaller, both due to fill-rate losses and to angle inefficiencies. You need to make a photo cell a LOT better to keep performance at a constant level while you shrink cell dimensions.
The angle problem is starting to show up in both the 50D and the 7D, Qe goes down noticeably as you deviate from a perfect 90º light incident angle. This does not only affect wide-angles - give it some thought! An F/1.4 lens has an aperture that is 1.4x smaller than the focal length. If we assume a "normal" lens construction, that means that F/1.4 gives light angles that hit the sensor surface from a [atan(0.5*(1/1.4))] 20º angle. F/2 is at [atan(0.5*(1/2))] a 14º angle. This is the part of the picture that's smack in the middle of the center, out towards the edges you have to add the chief-ray angles, which in a normal 50mm lens is 17º.
Efficiency when used with large aperture lenses is lowered, and vignetting increases - this should be accounted for when you calculate "average Qe" also - as it has a very real effect on a "normal picture", not just in a test-bench.
Leica found this out the hard way when choosing the Kodak 6.8um cells for both the M8 and the M9. Symmetrical wide-angles give VERY high angles of incidence on the sensor, and hence they have 2-3Ev worst-case vignetting coupled with very noticeable hue-shifts towards the edges of the frame - not to mention a 60% efficiency loss with all short large aperture lenses....
Interesting info, Suede. This had me puzzled though:
theSuede wrote:
Effective Qe in a real situation always goes DOWN as the pixels get smaller, both due to fill-rate losses and to angle inefficiencies. You need to make a photo cell a LOT better to keep performance at a constant level while you shrink cell dimensions.
I understand the bit about fill factor etc, but the LX3 (2µ pixels) tests out to about the same number of photons per RAW level per square micron as the D3 (8.45µ pixels). Granted one is CCD and the other CMOS, and the digicam has no AA filter, but it indicates that the losses might not be so great as the pixels shrink.
Please don't forget that the LX-3 sensor is manufactured in equipment that DSLR-sensors can only DREAM about, and then add in that surface efficiency is a lot higher in those manufacturing lines. We just hit 180nm in the latest sensors (1D4, 7D, the latest D3s and panasonics latest) - that's the same level of manufacturing resolution as the small sensor business have had available for 6-7 years now. We just saw the first sensors incorporating Cu leads in stead of Al.
The 7D sensor manufactured in LX3-class specifications would quite probably make the 7D more expensive than the 1Ds3. Cost = area squared divided by manufacturing technology maturity.
Surface efficiency is very important, as silicon crystal has a very high refractive index - 3.6 something for average green light - compared to "high" refraction special glasses used in lenses - normally around 1.8. So, as long as you can get the light to hit the sensor surface at the right place, light is bent to pass almost straight down into the sensor, no matter what the angle of incidence was. You have to get the light there in the first place - and this is a lot easier in a small die sensor. They have a lot thinner filter elements, and some other manufacturing advantages too that add to the surface efficiency.
Small-sensor applications almost never have to worry about angles either. Almost all used lenses are retrofocals of low angles - and apertures larger than F/2.8 are very rare. And when you design a fixed lens system, the lens and sensor can be optimized to the same incident angles - you always know what the sensor will be used with.
theSuede wrote:
We just hit 180nm in the latest sensors (1D4, 7D, the latest D3s and panasonics latest) - that's the same level of manufacturing resolution as the small sensor business have had available for 6-7 years now.
Very interesting info, Suede.
Btw, how do you know that Canon is using 180nm on the 7D and 1D4?
By assembling info from different places around the web, my impression is that Canon is still using 200nm process on 200mm wafers.
Chipworks calls the SX1 sensor revolutionary (for Canon, that is).
I'm wondering if DSLR sensors will be manufactured soon using the hybrid process - or this technology will make them super expensive, as you said, and will not be viable for them.
Sorry, a very "normal" fault of mine. There is no such thing as "know" in this business, we don't even "know" what plant (or combination of plants) was used for some of the sensors.
What I DO know is that some 180nm machinery moved on in (was scrapped from "serious money use") into sensor fab around the time when the 7D and 1D4 sensors realistically should have started beta-production. I also "know" that euipment now capable of running FF-sensors without stitching is available to the imaging group subdivision of Canon.
This does of course not mean that I KNOW that they are being used at full capacity, and I should not express it to make it sound that way.
The BEOL-view of the 1.75u sensor in the chipworks link is interesting - compare the scales of the cell with a larger cell sensor... They're not that different! the ratio of cell width to package layer height is almost the same - which explains most of why the small sensors can be made so light sensitive. It also explains why their colour discrimination capability is no better at ISO80 than some FF/FX-cameras can manage at ISO3200... :-) Are we willing to let our cameras loose 90% of their colour resolution capability for "high-resolution" purposes?
( I normally shoot with a 1Ds-II
am I doing something wrong ? 2.8, 1/13, ISO 400 on a tripod...70-200 IS USM
