I really do think that the majority of reports of softness from the 7D aren't the camera at all. Of course smaller pixels demand better lenses and technique if you want 100% crops to look equally sharp. If people uprezzed their 40D images to 18 Mpixels AS THEY SHOULD when comparing against the 7D, this controversy wouldn't exist, and the 7D would almost never look softer than the 40D.
This issue arises every time we get a large bump in sensor resolution and people discover that their lenses aren't good enough, or their aperture choices weren't optimal, or their shutter speeds weren't fast enough to max out the camera's available detail. The camera gets blamed because they zoom into 100% and see 'softness'.
If you want to use the full 18 Mpixel resolution of the 7D you need to think carefully about lens selection. Some really excellent lenses won't give crisp pixels at 7D resolution with many aperture settings.
I predicted this controversy when the 5D2 was due, and with the 7D it's the same controversy with knobs on since the 7D is the reachiest DSLR ever:
This is yet another side-effect of the equal magnification problem that I am trying to turn into a religion. If people won't compare camera crops fairly, at equal physical image size, then they should just shut up, listen, and learn....Show more →
on top of that DPP seems to get a bit soft with the same settings and ACR can be a trace less crisp at times at 100% view, probably due to counteracting mazing, but if you step back and just look at total detail captured, it is definitely more than than 20D/40D/D300s/etc. and even a trace more than 50D and the image isn't as jaggy looking.
The 7D has made a large impact on the DSLR market compared to the D300s. Look at the reviews on B&H. The 7D has 89-100 reviews (the search page says one thing, and the product page says another), but the D300s only has 25. And the D300s was released earlier. When I tried it, I loved the feel of the controls, and the VF was fantastic. It was speedy, and had a solid feel.
It has also made an impact on the video market. The 7D and 5D Mark II have been seen shooting footage more and more.
I think Canon has a keeper here, especially if they can fix the issues some people say they have.
It's nice that so many spend so much time trying to understand complex equipment. And ultimately it doesn't matter to me, if I'm not happy with the equipment. Anybody can show on a piece of paper anything they want in theory. If it came down to the fanboys with their graphs and charts and smugness telling how I should feel about a camera I'm sure I'd be set here.
But I simply posted, and I made sure I said this plainly "non-scientific" post of how I PERSONALLY feel about the 7D at this time and how it pertains to my shooting.
Naturally the first thing I get is people like you trying to be condescending that it must be entirely my fault I am not as happy with the 7D as they with their charts and stuff think I should be. My reasonable complaint is I am not satisfied in many ways with the end result of photos the 7D gives me. Something is "not quite right" in too many cases. I don't need a PHD to know when I don't like something just my eyes. Nor is my dislike of the results any less valid because you think it so.
I will give you that this camera definably drives home the need for better glass to get all you can out of the camera. Which unfortunately is not always a good thing in my opinion. Very rarely before have I felt that I needed IS so much in my lenses as I do with this camera.
So there.
I don't care if you are sympathetic, I do L. H. Smith wrote:
Considering how much time and effort many put in here trying to understand how to use this complex equipment for best results, and considering how much time and effort others put in trying to teach/help with that goal -- sometimes it is just plain difficult to be sympathetic with certain posts.
1."There is a lot I don't know or understand about this equipment and its use."
2. "I don't care about the info available to explain it to me and help me learn."
There are reasons for the different categories of photo gear (casual user/serious amateur/pro).
There are also levels of knowledge, technical and other, appropriate to and required for each category.
When learning more seems too great a challenge, a person may have reached his ceiling.
A bicycle will get you to the store for the loaf of bread at 10 mph easily and simply enough.
With a bit of coaching or experimentation, even a cave man could do it. ;-)
A formula race car will get you to the store at 10 mph also.
But if that is all you need to do, using the race car would hardly be worth the expense and the dreaded "learning curve".
One thing for sure, the caveman who "doesn't care" about any "explanations" would probably be best advised to stay with the bicycle.
If "I take photos that people pay me for" is good enough, and the philosophy is "increased knowledge is not for me", then perhaps where one is will be where one stays.
A perfectly fine choice, if someone thinks so, ...but where is the basis for *reasonable* complaint?
Saturos, I don't expect you to believe charts and graphs from fanboys. I'm certainly not a fanboy touting a chart or a graph.
If you believed the "logic" of some of these imbeciles, the mindset would be that less pixels=better, and that's that. By that logic a Canon Powershot G2 would be better than a G10, and the 5D original better than the 1dsmk3 (purely from an IQ standpoint).
Those DxO ratings to me are bull, and to be taken with a grain of salt. Pretty much just something that you would read if you were taking a poop.
What I would expect you to believe is real world results, or a real test (not a graph test or some crap). I clearly have a 7D that outperforms my 40D and have photos to prove it.
If you have a problem with your 7D, I suggest returning it or sending it to Canon, because you should be happier than a pig in poop about it.
on the above
1.I dont even understand any of the charts..then again why should I?
2.like in HDTVs 1080p has more pixels than 480p,but the 1080p looks better wtf
3.while taking a poop prefer to read my Canon EOS manual which everyone should before posting troll stuff like the 7D sucks DxO said so...
IMHO more pixels is better,that tired old argument BS given the new technology in the 7D ...I hope you post more pic Robert at least I will be able to know how to judge the 7D doing what it is supposed to do... take pictures
brainiac wrote:
This is yet another side-effect of the equal magnification problem that I am trying to turn into a religion. If people won't compare camera crops fairly, at equal physical image size, then they should just shut up, listen, and learn.
Amen, brother!
I have a vision ... that all cameras will be compared equally(at equal image size) ...
First off, let me just say technical mumbo-jumbo is cool, I dig it, but it's meaningless if your camera and/or lens is a loser. Trust me, it doesn't take rocket science, charts and/or graphs to determine that one lens doesn't perform well as compared to another ... ditto for camera bodies.
Assuming you are "absolutely positive" that apples are truly compared to apples (as versus oranges) and have arrived at a conclusion about your gear then go with it. If you have any concerns regarding your test approach then pay attention to some of the suggestions in this thread and make sure your final decision is based on apples-to-apples.
If you are questioning the AF performance of a lens then test it on many bodies and then make a decision (or find many like lenses and test them on your single body). If you don't have many bodies or friends with cameras then try your local photography store.
If you are questioning the AF performance of a body then test it using many lenses and then make a decision (or find many like bodies and test them using your single lens). If you don't have many lenses or friends with lenses then again, try your local photography store.
If you are questioning the IQ of two equally sized sensors (equal physical size) of different resolution then make sure you normalize the files via interpolation to equalize the size of the test object before making an evaluation using your monitor.
If you don't understand the how-and-why associated with normalization of two files from two different cameras prior to viewing them on a monitor then make two equal sized prints and perform your visual evaluation.
Oh, one more thing, if you want to utilize DxO minus much self-confusion and/or arguments from other folks then just stick to the SNR 18% and Dynamic Range comparisons ... but be absolutely certain to click on the "Print" button (as versus "Screen") prior to your evaluation. If SNR 18% and Dynamic Range mean absolutely nothing to you then trust me, you will be better by avoiding the site.
jkurkjia wrote:
Oh, one more thing, if you want to utilize DxO minus much self-confusion and/or arguments from other folks then just stick to the SNR 18% and Dynamic Range comparisons ...
I used to do that. But after I realize that even their own engineers do not trust their own results, I have given up on them.
PS: If you do not know what I mean, just look at the SNR18% and DR results under the Print tabs for 7D vs 50D. Then, look at the scores they assign for low light ISO. Shocking, no?
PS2: It's also hard to believe that site anymore when their results directly contradict those 7D vs 50D RAW pics posted in the-digital-picture, DPReview, Imaging Resource etc etc.
digitalbug30d wrote:
on the above
1.I dont even understand any of the charts..then again why should I?
2.like in HDTVs 1080p has more pixels than 480p,but the 1080p looks better wtf
3.while taking a poop prefer to read my Canon EOS manual which everyone should before posting troll stuff like the 7D sucks DxO said so...
saturos wrote:
But I simply posted, and I made sure I said this plainly "non-scientific" post of how I PERSONALLY feel about the 7D at this time and how it pertains to my shooting.
If a tool is no use for you and you personally don't find any value in it, by all mean sell it, if that makes you happier. Not all tools are meant for everyone.
thw2 wrote:
I used to do that. But after I realize that even their own engineers do not trust their own results, I have given up on them.
PS: If you do not know what I mean, just look at the SNR18% and DR results under the Print tabs for 7D vs 50D. Then, look at the scores they assign for low light ISO. Shocking, no?
PS2: It's also hard to believe that site anymore when their results directly contradict those 7D vs 50D RAW pics posted in the-digital-picture, DPReview, Imaging Resource etc etc.
maybe the better color sensitivity at high iso explains the difference in your PS?
skibum5 wrote:
maybe the better color sensitivity at high iso explains the difference in your PS?
I think the main point to be drawn from this is that DxO's 'scientific' method of assessing sensors is considerably less informative than showing actual results taken with the cameras in carefully described circumstances. The overall scores are entirely arbitrary. The SNR stats don't offer insight into noise quality, which is very important for anybody who struggles with low-light noise. The icing on the cake is that DxO prefers to mislead by showing the per pixel stats as default, when they have even published a page read by few of their visitors which states that the per pixel graphs are basically nonsense.
The sooner sites like DxO, DPR, and LL grasp this simple scientific point, the sooner their visitors can put a little trust in their posted examples and make informed choices.
theSuede wrote:
Alundeb: Spatially dependent noise sources (banding) affect low ISOs more than high ISOs, even at 18% saturation. It DOES have a very big effect on the figures you get - try it and see. The banding effect compared to total noise then gradually diminishes until it's almost non-existent at ISO12800, even in very "bandy" testsubjects. The slope of the 18% SNR is -2.6Ev per doubling at DxO, which is quite exactly as expected, if you have a spatially dependent noise-source present.
Measured with frame-subtraction, the numbers are VERY linear at -2.9dB per doubling of ISO.
I just checked the vertical banding in the uniform patches of some of my early measurements of the 7D, and found this:
The amount of banding is largely independent of ISO, and increases linearly with exposure. It is clearly a gain variation, not offset. This does not apply to the banding found in black frames, though.
After row averaging of CFA split raw data, on my copy, the banding amounts to a standard deviation of about 9 ADU at 18%.
Assuming banding does not add in quadrature to random noise, but close to linearly, it contributes 2.35 dB to the measured noise at ISO 100, 1.7 dB at ISO 200, 1.2 dB at ISO 400 and then gradually less. So this explains the slope at DxO, yes.
alundeb wrote:
I just checked the vertical banding in the uniform patches of some of my early measurements of the 7D, and found this:
The amount of banding is largely independent of ISO, and increases linearly with exposure. It is clearly a gain variation, not offset. This does not apply to the banding found in black frames, though.
After row averaging of CFA split raw data, on my copy, the banding amounts to a standard deviation of about 9 ADU at 18%.
Assuming banding does not add in quadrature to random noise, but close to linearly, it contributes 2.35 dB to the measured noise at ISO 100, 1.7 dB at ISO 200, 1.2 dB at ISO 400 and then gradually less. So this explains the slope at DxO, yes. ...Show more →
I would be very curious to know whether the banding could be detected from column averages of raw data in typical images -- in other words, if one averages (or totals) all the pixel values in each column, is the column-to-column variation in the average or total reflective of the gain variation, or is it swamped by the column-to-column variation in the image itself. If not overwhelmed by the image, one could use this column average on the fly as a rough means of compensating the gain variation.
ejmartin wrote:
I would be very curious to know whether the banding could be detected from column averages of raw data in typical images -- in other words, if one averages (or totals) all the pixel values in each column, is the column-to-column variation in the average or total reflective of the gain variation, or is it swamped by the column-to-column variation in the image itself. If not overwhelmed by the image, one could use this column average on the fly as a rough means of compensating the gain variation.
Considering that NIK Dfine 2.0 performs debanding successfully on JPG images from my 7D, it should be possible with an even better approach in RAW.
High pass filtering and min/max limiting of the column averages would at least be necessary.
Since frame subtraction seems to work for theSuede, I would like to know myself if the gain fluctuations really can be static, or how much of them. It doesn't seem so obvoius in my case.
It must be said for the record that the vertical gain banding is always 47 dB below the signal, so the effect is very small.
