March 03, 2015, 12:16:29 AM

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - jrista

Pages: 1 ... 77 78 [79] 80 81 ... 334
1171
............
 I've long held the opinion that crop sensor cameras, like the 7D, do have value in certain circumstances. The most significant use case where a camera like the 7D really shows it's edge over full frame cameras is in reach-limited situations.

............

I'd like to prove my case

................

Both images were initially scaled to approximately 1/4 their original size (770x770 pixels, to be exact).

The 5D III image was then layered onto the 7D image, and upsampled in Photoshop by a scale factor of exactly 161.32359522807342533660887502944%.



While I agree with you that a 7D (or any so called crop sensor) can have advantages over a so called full frame sensor. I think you need to review your work if your objective is to reach a valid conclusion.

1) you start off with a strong opinion. (its better to have an open mind)
2) Then you try to "prove my point". (it might be better to try to test your opinion)
3) Then you do something that is going to be very detrimental to one of the images.

You may claim that you would have to upsample the 5D3 image to get the same size as the 7D. But you have already down sampled it - so you have lost detail in the 5D3 file.

To demonstrate I made a simple file in Photoshop. 770 pixies  ;D wide, copied it, scaled it to 481 wide, then upscaled it to 770 wide. Hardly by chance my file had two types of detail. A sharp line and a not so sharp line. The result can be seen below.

I think I have just proved that photoshop is better than photoshop.  >:(

Don't get me wrong I'm a fan of the 7D and think its a great camera. I also think there is a place for "crop sensors". I'm waiting for the 7D2, I don't think it will be for me, but I definitely see a crop sensor shaped hole in my kit.

and finally whats with 30 odd decimal places!


EDIT: Just in case anyone wonders  ;D  the down sampling and up sampling were done with default PS settings

You are correct. However, the image below was actually done a bit differently. In this case, both samples were downscaled to fit in the 770x770 pixel image...the 5D III image was not first downsampled then upsampled again.



Your right, certainly not as stark a difference as my first example. Maybe that one is invalid. This example, however, does show that the 7D is still picking up more subtle details and nuances of color. The differences are not stark, but they do exist. Also note, both of these images were denoised. They were both denoised to the point where they both exhibited about the same noise levels...where noise was pretty much not visible. Obviously, there was quite a bit less noise reduction applied to the 5D III image.. That actually costs the 7D a little bit of it's detail as well...but it is on a level playing field with  the 5D III as far as noise goes, so I still think it's a fair example.

1172
Very nice bird photos.I know this has been hotly debated here but I am pretty sure its the pixel size and not the sensor area that affects noise the most. My reasoning is that I have looked at images from the d800 which has aps-c size pixels and exhibits aps-c levels of noise when viewed  on a per pixel basis. That tells me that pixels of a similar size in the same generation show similar noise levels?

Your talking about on a per-pixel basis. On a per-pixel basis, that is true. However I'm talking about on a whole-image basis, or as it's called, on a "normalized" basis. When you compare images as a whole at the same size, assuming the same absolute area of sensor was used, then there won't be any difference in noise regardless of pixel size. There will, however, be a difference in detail.

This all assumes same pixel generation. The 5D III does have an advantage in upsampling due to it's newer pixel generation. It has higher quantum efficiency and overall a better pixel architecture, than the 7D pixels. That means less noise per pixel. I actually wish I had a 70D. That would make for a better comparison, as then both cameras would use sensors of similar generation, instead of being separated by over three years of technology. That's unlikely to happen unless I meet someone with a 70D who will let me borrow it for a night, though...as I have no intention of buying a 70D.

1173
Reviews / Re: DxO reviews Sony A7s: king of low light photography?
« on: August 09, 2014, 04:03:09 PM »


Except...they store the output in a lossy image format. Until Sony fixes that, at the moment, I don't think there is any significant difference between the A7s and 1D X for still photography. The big differences show up with video...the A7s thanks to the BionzX processing, does indeed do wonders at ultra high ISO video.

I think you exaggerate the influence of the lossy compression. It creates some very visible artifacts in very specific cases when you push the files, in around 3800 images with my A7R I have seen the artifacts in exactly 1 image. In the rest I doubt you could tell which one is which had both the compressed image and the uncompressed been available.

But I think they should make a uncompressed option, which they could easily do with a firmware update. But the A7 cameras have also a lot of other silly firmware limitations and honestly I'd be surprised if Sony bothers fix them. There will never be a perfect camera.

It isn't the artifacts. It's the fact that the compression is lossy. They are throwing away information. It's the original depth of information that makes Sony Exmor so great. Why, then, would you throw away some of that extra precision? Sony cameras, when tested, don't achieve the same levels of DR as Nikon cameras with the same sensors...so clearly, Sony is losing something by using their compressed RAW format. Because it's lossy, it doesn't even qualify to be called RAW...it isn't RAW! :P

1174
I have been using macbook retina's for awhile strictly running windows since the screen covers the rgb gamut and is very accurate. I just bought a thinkpad w540 with the 3k display. It covers 97% so close enough but the pics just do not have that pop the had with the retina since the thinkpad has an antiglare coating. The pictures look flat and the mac makes the appear to have more saturatiin. Does anyone here edit with a matte coated screen thqt has worked with a macbook that could tell me your experience? I am afraid the macbook is the better screen but I never run osx.

You should look into the Dell XPS 15. I just purchased the XPS 15 with 3200x1800 screen, 500GB SSD, and 16GB ram. The screen is amazing. It had a slight greenish tint at first, but after calibration, it's wonderful. It has a reflective screen, which honestly I don't like that much...it glares all over the place, which is actually worse than having an antireflective screen.

I really wish that manufacturers would start using some kind of multicoating on glossy screens. That way we could have the best of both world...very low glare, but still have that crisp, crystal clear detail.

Anyway, the Dell XPS 15 is a phenomenal laptop. It's extremely fast, has enough memory to do just about everything, and the screen is very, very nice.

1175
I've said it so many times, I know others have also said it. Noise is relative to the area, not the pixel. If two cameras use the exact same area of sensor to resolve a subject, then there is no difference in noise. Not when the image from the sensor with smaller pixels is downsampled to the same dimensions as the image from the sensor with bigger pixels.

So then why do you keep questioning DxO sensor normalization? What you just described above is exactly the same thing. So why do you get it in one context, but not in the DxO context?

We've had this argument. We've had it countless times. You know the answer. I've been extremely detailed and clear about my opinions, and exactly what my opinions are. I'm NOT going to let you ruin this thread by diving into another pointless discussion of DXO. You want to have that debate, please, don't ruin my thread...start another thread. PLEASE.  >:(

Everyone, let's keep DXO out of this discussion. This thread has nothing to do with comparing Canon and Nikon cameras or anything like that. It has to do with the reach advantage of smaller pixels. That's it. I don't want this thread to be derailed by another useless debate that we've all had ten thousand times.

1176
Reviews / Re: DxO reviews Sony A7s: king of low light photography?
« on: August 09, 2014, 03:43:32 PM »
...
The Sony A7s is king of the hill at the moment when it comes to low light performance. It has better high ISO than anything else out there. On top of that it has class leading DR - not as good as Nikon but definitely better than anything Canon currently has to offer.

All that the A7s needs now is a reasonable selection of lenses.

A question that many will be asking is whether the A7s will be beaten on high ISO. That's hard to figure and quite possibly not, even by its successors - or at least for bayer sensors. The reason for that is fewer, larger, pixels and that is something that a lot of people have been screaming for.

"Give me fewer pixels but make them better (bigger)" is often heard in internet forums.

Well now Sony has done just that and hasn't it delivered.

Fewer, larger pixels don't have anything to do with it. Pixel size has never really had anything to do with it. Sensor size, quantum efficiency, and fill factor are what really affect it.

And how do you get more electrons in the pixel? By making it bigger.

More electrons per pixel is a facade. Let's say I have 10µm pixels and 5µm pixels. The large pixels gather 200ke- at FWC. The small pixels gather 50ke- at FWC. The bigger pixels win, right?

Sorry, wrong. Bin four 5µm pixels, and what do you get? 50ke- + 50ke- + 50ke- + 50ke-. That's 200ke-! Pixel size does not matter. Assuming you put the same ABSOLUTE SENSOR AREA onto your subject, it doesn't matter what size the pixels are...your still going to be gathering the same amount of light in total.

When you throw in fill factor, the 5µm pixels won't quite get 200ke- light in total. It might be something like 195ke- or so. That is still a very small difference, about 2.5%. Overall, smaller pixels still gather the same amount of light as larger pixels, if you put the same absolute sensor area on the subject.

Pixel size isn't what matters...the underlying technology is what matters. The efficiency of the sensor, the rate at which it accumulates dark current, the readout mechanism and how much downstream noise it introduces. THOSE are what matter. In that respect, Sony definitely has the better technology (no one questions that.)

Quote
I'm guessing the high low ISO read noise was some kind of tradeoff necessary to achieve the clean high ISO.

Or maybe not clean high ISO but how to get the data off the sensor fast enough to do 4k video. Whatever it is a decision that Sony made.

Indeed. Similar to the decisions Canon made when they decided to make the 1D X and 5D III excellent high ISO performers that are capable of handling very high still photography frame rates. The data throughput requirements to chomp through 14 18mp images per second is similar, if not higher than, the data throughput requirements to chomp through 4k video.

But, it's ok when Sony makes such a decision, and not when Canon does? (That's all I can gather from your history here on CR...)

Quote
Anyway, pixel size has little to do with it. Transistor size has a small impact...it affects fill factor, especially as pixel count goes up, however a relatively "low" pixel counts of 18mp and 12mp, the difference in transistor size is minor.

I don't know how to respond to this. It's all just wrong. Transistor size is related to the lithographic process used to create the sensor. Thus a 12MP and 18MP can have exactly the same size transistors whilst having different size pixels. Now there is a "fill factor", if you like, where space taken by a transistor and its traces is not available to a pixel.

But since you've raised the topic of transistors, yes, Canon still has larger transistors than Sony and that's not a good thing.

No, Canon's larger transistors are not a good thing. I totally agree they need to improve their manufacturing process. Actually, they already have...Canon has been using a 180nm copper interlink design with lightpipes for their small form factor sensors for years now. Who knows why they haven't employed it on their larger sensors.

As for fill factor, the difference that larger or smaller transistors make with such large pixels is small. It's there, certainly, but it is not the most significant factor. It doesn't come even remotely close to the 20% difference in quantum efficiency. THAT is a truly major breakthrough, to achieve 67% Q.E. at room temperatures. It's the increase in quantum efficiency that really gives the A7s the edge here. Combine 67% Q.E. with Sony's better readout system, and yea, it's going to be a hell of a performer.

Except...they store the output in a lossy image format. Until Sony fixes that, at the moment, I don't think there is any significant difference between the A7s and 1D X for still photography. The big differences show up with video...the A7s thanks to the BionzX processing, does indeed do wonders at ultra high ISO video.

1177
Alrighty. I've updated the original post with more examples. I've added full 1:1 crop comparisons between the 7D and 5D III, WITHOUT scaling the 5D III. That should clearly show how much smaller in the frame the moon is with the 5D III, at exactly the same focal length.

I've also added a noise comparison example, which shows both images at native size, then downsamples a 7D of the same region to 5D III dimensions.

1178
Which has greater noise? An APS-C sensor or a full frame sensor cropped to APS-C size? Bare in mind our hypothetical situation is you're still reach limited, so the bigger sensor in itself conveys no advantage, and the only arguable difference is pixel size. For roughly comparable sensor generations I'd argue they're practically the same. Outside of lab tests, it probably isn't significant.

At ISO6400, I'd happily use either of my 600D or a 5D mk2 (as secondary body to 7D), but when reach limited the 600D would be my preference of the two. To me noise isn't the limiting factor in this scenario.

Below are a pair of images shown at 100%.   One is from an 18 MP APS-C camera at ISO 3200.  The other is from an 18 MP FF camera at ISO 6400, a full stop higher than the APS-C image.   

I'm having trouble telling which is which, the noise levels are so similar.   ::) ::)



I think the 7D can do even better than your example. Here is a bird photographed with a 500mm f/4 L II:



Very low noise. Here is another:



Also very low noise. The 7D, when used properly, can be a truly superb camera. I think people get caught up in the noise levels when they first use it, then make a decision early on that the 7D simply cannot produce low noise results.

I'm a very well versed guy when it comes to photography. I do not have pro-level skill and my images don't exhibit pro-level quality, but that is simply a matter of practice. I still have to work, and I work my butt off to pay for the kind of equipment I buy. I know what the difference between the 5D III and 7D is. I've had more than enough time with the 5D III, between playing with other peoples out in the field, to having had mine for several months now. Things are what they are...6.25µm pixels vs. 4.3µm pixels. Smaller pixels mean more detail. Greater sensor area on subject means more light. Normally, a full frame sensor is capable of putting MORE sensor area on subject...however that is not the case in a reach-limited scenario. In a reach limited scenario, the same sensor area is on subject. That means the only significant difference is pixel size.

I've said it so many times, I know others have also said it. Noise is relative to the area, not the pixel. If two cameras use the exact same area of sensor to resolve a subject, then there is no difference in noise. Not when the image from the sensor with smaller pixels is downsampled to the same dimensions as the image from the sensor with bigger pixels.

1179
Nice comparison. Thanks for posting.
They were both shot with the same settings and the same processing in Lightroom and I wonder what happens if you use the best settings and most optimal processing for each camera/image.

With more than two and a half times more light, it's two and a half times better. Like using two and a half stops lower ISO on the cropped sensor.

Double the light is one stop so 2.6 times the light is about one and a quarter stops.

You are correct about the number of stops. My mistake.

As for settings...what would be better settings? I mean, exposure is exposure...and technically speaking, using the same ISO means the 5D III has the advantage, no? I used ISO 200 for both shots...the larger pixels of the 5D III should mean that much more light is gathered per pixel at ISO 200 (which is indeed the case, noise comparison coming.)

So, I honestly don't think I could have used any better settings on the 5D III. And exposure is exposure...it's light over time...for a given subject of given brightness, you have to use the same exposure.

1180
Jrista,

Do you think you could post the unprocessed RAW files?  I'd like to see these with no noise reduction. It almost looks like the 5DIII had too much applied or was out of focus.

I'll post some 100% size crops (no scaling) so you guys can see the noise.

Regarding the foucs...that is actually the impact of seeing. As I described in my original post, with a longer focal length, you can kind of "cut through" that a bit more. Maybe that is skewing the results a little...but remember that both images are suffering from the effects of seeing, not just the 5D III. The 5D III suffers a little more from seeing, however I did choose the best out of five of both the 5D III and 7D images.

The thing about seeing is there are moments when the turbulence "flattens", and things get really clear. There is actually an astrophotography technique, used even by professional observatories, called "Lucky Imaging". Described in 1978 by David L. Fried, the technique involves taking many (sometimes many thousands) of exposures of the same celestial subject in sequence. Over the span of those exposures, many will have near-perfect seeing, and the subject will show up clearly. Professional observatories might use highly sensitive imagers to take tens of thousands of short exposures of a single subject, then pick the best 500 and integrate those to get extremely clear results that rival Hubble images (which does not suffer from seeing at all, given it's in space). Now, I did not take thousands of exposures.

Something I can certainly try is to set up on a better night with better moon features, and take several hundred each with both cameras, and pick the best out of those frames. That is a much more significant task, however I'm happy to do it if it would help further prove the point. I do not believe it will change the results. There are 2.6x more pixels in the area of the moon for the 7D than the 5D III. If we have 3 million pixels on subject with the 5D III, we are going to have 7.8 million pixels on subject with the 7D.

It really doesn't matter either way...the 7D is ultimately going to win the detail battle when your reach is limited.  I'll share more images, and I'll see if I can get the RAWs uploaded somewhere so you guys can experiment yourself.

1181
Both images were initially scaled to approximately 1/4 their original size (770x770 pixels, to be exact).

The 5D III image was then layered onto the 7D image, and upsampled in Photoshop by a scale factor of exactly 161.32359522807342533660887502944%. This scale factor was derived by computing the sensor diagonals of both cameras:

That would give an unfair advantage to the 5D III. You have to do exactly the same things to both images to keep both results consistent. Even if I did not crop at all, the 5D III image would still have to be upscaled by the same amount.

I performed every action identically, for both the 1/4 scale and 1/2 scale images, then added the one extra step of upscaling the 5D III moon to the size of the 7D moon. That was simply to normalize subject size.

I can do it again, and leave everything at 100% scale. The moon in the 5D III shot is MUCH smaller than the moon in the 7D shot, so it really doesn't matter if I scale first or not. Here is another example...these are cropped and scaled to 770px, but the 5D III image was not upscaled...it's at it's native size:



For reference, these are full size crops...the full height of the sensor is used...I only cropped out empty black space to the left and right. For all intents and purposes, these are downsampled 1:1's. I'll get some more images of them at 100% size, without any noise reduction.

1182
Photography Technique / Re: Questions about Shooting the Supermoon
« on: August 09, 2014, 03:31:51 AM »
I did some moon imaging tonight at 1200mm, with the 5D III and 7D. Here is a high res (superresolution or "drizzle") of the last bit of shaded moon detail along the eastern edge. This was integrated in AutoStakkert! 2, from a 1000-frame video taken with my 7D at 1/100s f/8 ISO 200:




I could have probably gotten even more detail, but I had BackyardEOS configured incorrectly. It was only doing 5x zoom in the planetary imaging mode, when it can actually do up to 10x zoom. I ran out of time, the moon slipped behind the trees. I'll try again once the moon is in more of a waning gibbous phase, when it should have a lot more interesting detail with some shading.

Here are some full frames, comparing the resolving power of the 7D and 5D III:




1183
Third Party Manufacturers / Re: DXO uh-oh?
« on: August 09, 2014, 03:16:01 AM »
The reason I prefer the joystick is b/c the D-pad requires too much travel for my thumb to change focus point. But neither the D-pad nor joystick are ideal. I can think of a much better way to select AF points quickly... can you? :)

But given how spectacular the D810 is at tracking the subject across the frame in '3D' tracking mode, the joystick vs. D-pad debate is less of a concern for me as I'm jumping ship. However, I'd still prefer a faster way to select AF point for those situations where AF tracking fails - e.g. in very low light, low contrast subjects, heavily backlit subjects, etc.

Interesting about the joystick breaking. Ultimately I don't care - that can be fixed so I'd prefer function over longevity. But that's just me.

I really wonder why Canon doesn't bring eye control back. It certainly seems like fans of the EOS 3 ECF really want it. It would be interesting to have whatever it is your looking at in the VF be focused...that would just rock. I am guessing the system was expensive, at least that's what I'd read in the past...in a film camera, it was probably one of the most expensive things. However in a DSLR, it's just one more expensive thing to add to the mix...maybe it pushes cost over the edge.

1184
Reviews / Re: DxO reviews Sony A7s: king of low light photography?
« on: August 09, 2014, 03:13:23 AM »
...
So, the thing you might be missing...is that HDR isn't difficult these days. If you need more DR, take an extra frame or two (or 14). That works in every camera, regardless of it's sensor capabilities.
...

HDR is only useful where you can use a shutter speed of > 1 second because otherwise the inter-frame changes make it look crap. Think wind blowing in trees, moving water, etc.

That assumes landscapes. There are plenty of still-life use cases...such as PBD's interior design scene, or my description of the WWII plane interior. You could expose for as long as you wanted with such a scene, as there are no trees, or flowing water, etc. Most of the 15-frame HDR images I've seen were still scenes, usually interiors of something or some kind.

1185
Reviews / Re: DxO reviews Sony A7s: king of low light photography?
« on: August 09, 2014, 03:11:41 AM »
...
This thread really needs to be closed.

I thought about that during the week (I started this thread) but I didn't want anyone to accuse me of trying to censorship.

The Sony A7s is king of the hill at the moment when it comes to low light performance. It has better high ISO than anything else out there. On top of that it has class leading DR - not as good as Nikon but definitely better than anything Canon currently has to offer.

All that the A7s needs now is a reasonable selection of lenses.

A question that many will be asking is whether the A7s will be beaten on high ISO. That's hard to figure and quite possibly not, even by its successors - or at least for bayer sensors. The reason for that is fewer, larger, pixels and that is something that a lot of people have been screaming for.

"Give me fewer pixels but make them better (bigger)" is often heard in internet forums.

Well now Sony has done just that and hasn't it delivered.

Fewer, larger pixels don't have anything to do with it. Pixel size has never really had anything to do with it. Sensor size, quantum efficiency, and fill factor are what really affect it.

BTW, The A7s has the highest selectable ISO setting...however I've seen little difference between still photography images from the A7s and 1D X at ultra high native ISO settings. The Sony sensor technology is better. They have higher Q.E. by a pretty significant margin...67% vs. 47%, a 20% edge to the A7s, and they use a smaller process which means a bit higher fill factor. But then they go and gimp it with a crappy lossy RAW file format.

Another interesting thing is that unlike other Sony Exmor sensors, the A7s actually has high read noise. It's 21.7e- at ISO 100, where as the A7r, like all other Exmor sensors, is a much lower 5.2e-. The A7s does much heavier image processing with it's BionzX processor, maybe that's a source of additional noise. Whatever causes it, it's costing the A7s low ISO IQ...it has the potential to utterly trounce even the D810. If they could get the read noise at ISO 100 down to 3e-, they would have 15.7 stops of native sensor DR. If they then moved to 16-bit CP-ADC, their sensors would have a 2 stop edge over the D810. I'm guessing the high low ISO read noise was some kind of tradeoff necessary to achieve the clean high ISO. Makes me curious....is Canon making a similar tradeoff with their sensors? Their high ISO is very clean in my experience...I shoot at ISO 12800 with the 5D III all the time, and it's amazingly clean. And the 5D III has around 30e- of ISO 100 read noise...

Anyway, pixel size has little to do with it. Transistor size has a small impact...it affects fill factor, especially as pixel count goes up, however a relatively "low" pixel counts of 18mp and 12mp, the difference in transistor size is minor. The most significant difference between the A7s sensor and the 1D X sensor is the quantum efficiency. A 20% benefit in Q.E. goes to the A7s...that is a very significant difference. At high ISO, it is very clear that the BionzX chip is kicking in and performing some significant post-ADC noise reduction, which is why ISO 400k is possible. In the long run, that kind of quantum efficiency could evolve into something much more significant, if Sony can figure out how to properly utilize their own technology (i.e. don't save it lossy!!)

Canon has similar noise reduction technology in their DIGIC 6 chip. It would be interesting to see if they employ that in a future DSLR. I am not sure if Canon can catch up to the 20% lead Sony has in quantum efficiency...that is a very significant gap. I've never seen Canon improve Q.E. more than a few percent between generations, but they would need to do something fairly dramatic to offset that edge that Sony has. Assuming Sony sells the sensor and BionzX processor they use in the A7s, I would be willing to bet a third party manufacturer could make FAR better use of it than Sony has...Sony just needs to get their heads out of the sand on their RAW format...

Pages: 1 ... 77 78 [79] 80 81 ... 334