Sony A7s II - THE DSLR Killer?

[jrista]

From a geometrical standpoint it (BSI) is certainly a question of diminishing returns as pixel size increases. Whether it's negligible remains to be seen, but as I read the following, it appears Sony management thinks they've reached the point at which pixel area is irrelevant. I suspect he's just blowing gas, since physics would seek to contradict him, but maybe that's their position, and any improvements will come from deeper wells.

From what I have read about Sony's BSI technology, it is about more than simply increasing photodiode area. The design of putting all the logic and wiring on the opposite side of the photodiodes, and having more room within which to layout all that circuitry, has apparently allowed them to improve the electronics as well, signaling and current & voltage levels and things like that, making the electronic signal that drives the sensor less noisy. The noisiness of the signals that drive the electronics affect noise in the image as well, so BSI is important for other reasons than increasing light sensitivity. Apparently, it also improves SNR by lowering noise.

 

[3kramd5]

From a geometrical standpoint it (BSI) is certainly a question of diminishing returns as pixel size increases. Whether it's negligible remains to be seen, but as I read the following, it appears Sony management thinks they've reached the point at which pixel area is irrelevant. I suspect he's just blowing gas, since physics would seek to contradict him, but maybe that's their position, and any improvements will come from deeper wells.

From what I have read about Sony's BSI technology, it is about more than simply increasing photodiode area. The design of putting all the logic and wiring on the opposite side of the photodiodes, and having more room within which to layout all that circuitry, has apparently allowed them to improve the electronics as well, signaling and current & voltage levels and things like that, making the electronic signal that drives the sensor less noisy. The noisiness of the signals that drive the electronics affect noise in the image as well, so BSI is important for other reasons than increasing light sensitivity. Apparently, it also improves SNR by lowering noise.

Right, there is a lot more going on that *just* BSI. Still, his statement implies that Sony users need not trade sensitivity for resolution for sensors using this tech. I imagine a future product will prove him wrong.

My 5dsr is taking ages (or at least it feels like it) to come in so I considered cancelling and getting the a7rII.

I bet you'll get the 5DSR before I get my A7R2. When I pre ordered it on June 17, the confirmation (BH) said "New Item - Coming Soon", and the website estimated a ship date of July 29. It now says August. Sony's website says August 26, and others say August 30.

I updated my BH order, and the new confirmation changed the verbiage to "New Item, Available for pre-order 08/01/15." I don't know if that means they aren't taking *new* pre-orders until August.

 

[jrista]

From a geometrical standpoint it (BSI) is certainly a question of diminishing returns as pixel size increases. Whether it's negligible remains to be seen, but as I read the following, it appears Sony management thinks they've reached the point at which pixel area is irrelevant. I suspect he's just blowing gas, since physics would seek to contradict him, but maybe that's their position, and any improvements will come from deeper wells.

From what I have read about Sony's BSI technology, it is about more than simply increasing photodiode area. The design of putting all the logic and wiring on the opposite side of the photodiodes, and having more room within which to layout all that circuitry, has apparently allowed them to improve the electronics as well, signaling and current & voltage levels and things like that, making the electronic signal that drives the sensor less noisy. The noisiness of the signals that drive the electronics affect noise in the image as well, so BSI is important for other reasons than increasing light sensitivity. Apparently, it also improves SNR by lowering noise.

Right, there is a lot more going on that *just* BSI. Still, his statement implies that Sony users need not trade sensitivity for resolution for sensors using this tech. I imagine a future product will prove him wrong.

Once the entire sensor surface is light sensitive, then technically speaking he is correct. Fill factor is no longer a concern with BI designs. On a normalized basis, the full sensor surface is going to gather the same amount of light for a given subject framing, regardless of what size the pixels are. Now, if Sony manages to overcome noise scaling (which they seem to have done to a small degree so far in some of their cameras), then it is possible that larger pixels may have better DR and SNR (because they have a larger FWC). In that circumstance, larger pixels would still be better for low light photography.

If the A7s II hits with big pixels, and ~3-6e- RN at ISO 100, then I'll believe Sony has indeed overcome noise scaling issues, and the A7s II would probably be the best low light performer on the market. At the moment, the A7s has higher starting read noise, so I'm not sure that bigger pixels will actually perform better than smaller pixels on a normalized basis with a BSI design.

 

[3kramd5]

From a geometrical standpoint it (BSI) is certainly a question of diminishing returns as pixel size increases. Whether it's negligible remains to be seen, but as I read the following, it appears Sony management thinks they've reached the point at which pixel area is irrelevant. I suspect he's just blowing gas, since physics would seek to contradict him, but maybe that's their position, and any improvements will come from deeper wells.

From what I have read about Sony's BSI technology, it is about more than simply increasing photodiode area. The design of putting all the logic and wiring on the opposite side of the photodiodes, and having more room within which to layout all that circuitry, has apparently allowed them to improve the electronics as well, signaling and current & voltage levels and things like that, making the electronic signal that drives the sensor less noisy. The noisiness of the signals that drive the electronics affect noise in the image as well, so BSI is important for other reasons than increasing light sensitivity. Apparently, it also improves SNR by lowering noise.

Right, there is a lot more going on that *just* BSI. Still, his statement implies that Sony users need not trade sensitivity for resolution for sensors using this tech. I imagine a future product will prove him wrong.

Once the entire sensor surface is light sensitive, then technically speaking he is correct. Fill factor is no longer a concern with BI designs. On a normalized basis, the full sensor surface is going to gather the same amount of light for a given subject framing, regardless of what size the pixels are. Now, if Sony manages to overcome noise scaling (which they seem to have done to a small degree so far in some of their cameras), then it is possible that larger pixels may have better DR and SNR (because they have a larger FWC). In that circumstance, larger pixels would still be better for low light photography.

If the A7s II hits with big pixels, and ~3-6e- RN at ISO 100, then I'll believe Sony has indeed overcome noise scaling issues, and the A7s II would probably be the best low light performer on the market. At the moment, the A7s has higher starting read noise, so I'm not sure that bigger pixels will actually perform better than smaller pixels on a normalized basis with a BSI design.

BSI essentially maximizes fill factor, but it doesn't negate he charge capacity of a large pixel vs a small pixel of the same technology. In any case, future Sony products will show whether his statement is accurate or not. If there is a forthcoming product (A7S2 or whatever) which has a 135-format frame, similarly-designed BSI sensor with copper traces, less than 42MP, which offers more than 100k ISO, his statement is false. If I were a betting man, i would bet on that being the case. Time will tell. It doesn't matter much to me, but I bet an A7S2 with low RN BSI pixels and high framerate would be great for Astro.

Edit: they do seem to be marketing with the same notion. See attached: I read that as "you get resolution without compromising (sensitivity)." Perhaps there won't be a 7S sequel, or perhaps it won't have better sensitivity but rather better video (unbinned full frame 4k high framerate or whatever).

 

[benperrin]

Hi Ben, I took my 645z and A7ii to Iceland and put it through their respective paces. About halfway through the trip when I went to view an image on the A7 it started to flick through the images all by itself (suggesting water ingress) and the wheel at the back of the camera that can click left, right, up and down to confirm pressure no long does it on the left side.

Sure, not everyone is going to shoot waterfalls with it but it's not weather sealed. Lots of fabulous features on the camera that means it's worth keeping but is absolutely and without question NOT a professional build camera. The worst it was subjected to was heavy mist. I never changed the lens on it either.

Making a 42mp camera and targeting landscapers who shoot in all weathers is a bit silly imho. So I'm probably going to hold out for the A9 if it turns out to be a true rumor. It's not so much about the size of the body for me but how robust it is.

I'm ready to jump brands but I need something secure to jump to. Right now the Sony's fall over at the most important step for me, build quality, dual card slots and jog dial AF selection points. (rather than their current more convoluted way).

Thanks for the info. I'm not convinced yet that these round of Sony cameras are better than DSLR's just yet. The dual card slots is an issue for me but the rumour is that the a7rII can't use a single focus point when using a third party adapter with Canon lenses instead it has to use all the points (if that makes sense). Seems like there are compromises with either system. Also all the Sony marketing shots I've been able to find are at iso 100. I feel like I can't trust the Sony Artisans to be even slightly objective either. So at this point the 5dsr is ahead for me but I'll probably add an a7rII at some point.

 

[jrista]

BSI essentially maximizes fill factor, but it doesn't negate he charge capacity of a large pixel vs a small pixel of the same technology.

No, it does not negate the charge capacity differences. However, in a normalized comparison, charge capacities are moot anyway. If we assume we have two BSI sensors, FF, one with 10 micron pixels and the other with 5 micron pixels. Let's assume we have normal noise scaling as well, so 12e- RN for the 10 micrin and 6e- RN for the 5 micron. If we downsample the image from the sensor with smaller pixels to the same size as the image from the sensor with larger pixels, there will be ZERO difference in SNR.

Here's the proof. The formula for SNR:

SNR = (PixelSignal * AveragedPixelCount)/SQRT(AveragedPixelCount * (PixelSignal + ReadNoise^2))

If we apply this formula for our two sensors, we would have to assume that the 10 micron pixels get four times the signal. If we are talking 18% gray midtones, that may be 50ke- for the 10 micron pixels, which leaves us with 12.5ke- for the 5 micron pixels. Then we have:

SNRsmall = (12500e- * 4)/SQRT(4 * (12500e- + 6e-^2)) = 223.29:1 = 49.98dB
SNRlarge = (50000e- * 1)/SQRT(1 * (50000e- + 12e-^2)) = 223.29:1 = 46.98dB

For a subject, framed the same, exposed the same, at the same ISO, scaled to the same output magnification (larger image downsampled to the same size as smaller image)...there is no difference in the amount of light gathered. With an FSI sensor, you would have to knock off some percentage of the maximum signal, and factor in a small loss of light from vignetting in the physical pixel well from the wiring and transistors (even with a microlens). With BSI, you don't have those problems (you can have other problems, like color crosstalk, but that's different.) Without the fill factor issue, it doesn't matter if you have large or small pixels. The total SNR from both sensors on a normalized basis is the same.


In any case, future Sony products will show whether his statement is accurate or not. If there is a forthcoming product (A7S2 or whatever) which has a 135-format frame, similarly-designed BSI sensor with copper traces, less than 42MP, which offers more than 100k ISO, his statement is false. If I were a betting man, i would bet on that being the case. Time will tell. It doesn't matter much to me, but I bet an A7S2 with low RN BSI pixels and high framerate would be great for Astro.

Edit: they do seem to be marketing with the same notion. See attached: I read that as "you get resolution without compromising (sensitivity)." Perhaps there won't be a 7S sequel, or perhaps it won't have better sensitivity but rather better video (unbinned full frame 4k high framerate or whatever).
[/quote]

 

[3kramd5]

But in downsampling to normalize, you have specifically traded resolution for that SNR. Verifying that Sony has overcome pixel pitch when it comes to sensitivity must be done at native resolution.

 

[jrista]

But in downsampling to normalize, you have specifically traded resolution for that SNR. Verifying that Sony has overcome pixel pitch when it comes to sensitivity must be done at native resolution.

That's not actually true. You traded image dimensions for SNR. You have not, however, traded resolution. Resolution and image dimensions are not the same thing.

Have you ever downsampled a larger image to smaller dimensions? Blurring occurs at certain frequencies. In the case of a bayer array, it's going to occur at roughly 3x the pixel pitch. The amount of blurring that occurs in my hypothetical 10 micron sensor, which is greater than the amount of blurring that occurs in my hypothetical 5 micron sensor, is BAKED INTO the signal. You can try to deconvolve, but that will only get you so far. You sampled the original information at a lower level, and you can only recover so much. Information you never had in the first place...well, no amount of code will ever 'restore' that.

The higher resolution sensor, however, DID gather more information, at a finer scale. That finer scale is not really lost when you downsample. It's refactored, the greater amount of information is redistributed, but it is not lost. Compare the downsampled 5 micron image to the original 10 micron image...the SNRs will be the same, but the higher resolution sensor DID have more fine scale information, and that information is factored into the downsampled results. It WILL look sharper, clearer, and more detailed than the 10 micron image.

 

[3kramd5]

But in downsampling to normalize, you have specifically traded resolution for that SNR. Verifying that Sony has overcome pixel pitch when it comes to sensitivity must be done at native resolution.

That's not actually true. You traded image dimensions for SNR. You have not, however, traded resolution. Resolution and image dimensions are not the same thing.

Have you ever downsampled a larger image to smaller dimensions? Blurring occurs at certain frequencies. In the case of a bayer array, it's going to occur at roughly 3x the pixel pitch. The amount of blurring that occurs in my hypothetical 10 micron sensor, which is greater than the amount of blurring that occurs in my hypothetical 5 micron sensor, is BAKED INTO the signal. You can try to deconvolve, but that will only get you so far. You sampled the original information at a lower level, and you can only recover so much. Information you never had in the first place...well, no amount of code will ever 'restore' that.

The higher resolution sensor, however, DID gather more information, at a finer scale. That finer scale is not really lost when you downsample. It's refactored, the greater amount of information is redistributed, but it is not lost. Compare the downsampled 5 micron image to the original 10 micron image...the SNRs will be the same, but the higher resolution sensor DID have more fine scale information, and that information is factored into the downsampled results. It WILL look sharper, clearer, and more detailed than the 10 micron image.

But that's true of all sensors, from brand new full frame exmor R to decade-old canon 500nm FSI sensors. Sure, maybe that's what he was getting at, but in the context of promoting new tech that would be a fairly disingenuous.

 

[jrista]

But in downsampling to normalize, you have specifically traded resolution for that SNR. Verifying that Sony has overcome pixel pitch when it comes to sensitivity must be done at native resolution.

That's not actually true. You traded image dimensions for SNR. You have not, however, traded resolution. Resolution and image dimensions are not the same thing.

Have you ever downsampled a larger image to smaller dimensions? Blurring occurs at certain frequencies. In the case of a bayer array, it's going to occur at roughly 3x the pixel pitch. The amount of blurring that occurs in my hypothetical 10 micron sensor, which is greater than the amount of blurring that occurs in my hypothetical 5 micron sensor, is BAKED INTO the signal. You can try to deconvolve, but that will only get you so far. You sampled the original information at a lower level, and you can only recover so much. Information you never had in the first place...well, no amount of code will ever 'restore' that.

The higher resolution sensor, however, DID gather more information, at a finer scale. That finer scale is not really lost when you downsample. It's refactored, the greater amount of information is redistributed, but it is not lost. Compare the downsampled 5 micron image to the original 10 micron image...the SNRs will be the same, but the higher resolution sensor DID have more fine scale information, and that information is factored into the downsampled results. It WILL look sharper, clearer, and more detailed than the 10 micron image.

But that's true of all sensors, from brand new full frame exmor R to decade-old canon 500nm FSI sensors. Sure, maybe that's what he was getting at, but in the context of promoting new tech that would be a fairly disingenuous.

What? ???

Were talking about the benefits of a BSI sensor design in a full frame camera. The benefit is that it eliminates fill factor as an issue. With BSI, that is no longer a concern. Therefor, pixel size is effectively immaterial. If you want the sensitivity of a lower resolution sensor, no problem: downsample. If you want maximum resolution, no problem...you already have it.

What's the complaint here? Big pixels, small pixels...with BSI, normalized SNR is going to be the same regardless. Therefor, why would you ever opt for smaller pixels? They are less flexible. You cannot magically get more information out of the charge accumulated by larger pixels, however you can redistribute the information in smaller pixels to get more SNR.

The benefits are greater than that, though. You don't just gain die space for the light sensitive side. You also gain die space for the circuitry side. You don't have to pack all the circuitry into the borders between photodiodes. You can consume the entire die space. That opens up a whole lot more possibilities beyond simply greater sensitivity. Multiple pixel memories for increased dynamic range, per-pixel CDS or even per-pixel ADC, better wiring layouts to improve signaling efficiency and performance, who knows what else. There is absolutely zero reason not to use a BSI design if it's possible, regardless of the size of the sensor.

 

[3kramd5]

So with FSI sensors, is the only advantage of larger pixels the higher fill factor? In other words, is well capacity irrelevant? Will RN always scale directly with pixel size?

I know we were discussing BSI, but everything you've laid out with respect to downsampling and its affect on SNR applies to all sensors; it is not unique to BSI, whereas I read his statement as an advantage afforded by BSI, not merely by the maths of downsampling.

If he meant "if you want high resolution, here it is. If you want higher SNR, just downsample the image," is that not some Mayeda could also say of the 5DS, for example?

The benefits are greater than that, though. You don't just gain die space for the light sensitive side. You also gain die space for the circuitry side. You don't have to pack all the circuitry into the borders between photodiodes. You can consume the entire die space. That opens up a whole lot more possibilities beyond simply greater sensitivity. Multiple pixel memories for increased dynamic range, per-pixel CDS or even per-pixel ADC, better wiring layouts to improve signaling efficiency and performance, who knows what else. There is absolutely zero reason not to use a BSI design if it's possible, regardless of the size of the sensor.

Agreed, hence buying one sight unseen once a FF option surfaced

 

[jrista]

Well capacity is much less of an important factor than you may think. So long as noise does scale as expected (and, in some cases, it may scale better with larger pixels), then you wouldn't see any difference in dynamic range or SNR. The reason for that is smaller pixels are seeing a smaller portion of the subject. A 10 micron pixel may be resolving a 2 by 2 arcsecond area of your subject. A five micron pixel, on the other hand, would only be resolving a 1 by 1 arcsecond area of your subject. The smaller pixel has less area, and thus lower FWC...but it's also only getting a quarter of the light. The whole entire system scales...so, assuming equivalent technologies in both sensors, well capacity isn't really a factor.

Yes, downsampling does affect every image, even from FSI sensors. The key factor with BSI is that fill factor no longer applies. Canon cannot say that about the 5Ds...and worse, Canon's fabrication process is still using 500nm, so the fill factor concern is very significant the smaller Canon's pixels get. With the 5Ds, despite having nearly 4.1 micron pixels, may not even have 3 micron photodiodes because there would be at least a 1 micron border around them, the space consumed by 500nm wiring and transistors. Sony could have the same light gathering diode area with 3.36 micron FSI pixels, which in FF would be a 10000x6667 pixel 66.7mp monster. With BSI, the pixels would need to be marginally larger than 3 microns, which would be a whopping 12000x8000 pixel 96mp behemoth.

Because fill factor does not apply with BSI, big pixels or small pixels, it doesn't really matter. Everything scales, so your gathering the same total light, any given unit-area of sensor has the same photon flux, and dynamic range is the same on a normalized basis. There may be some slight real-world nuances that result in very small differences, but in general...there would be little reason to choose bigger pixels over small if you had two cameras to pick from and that was the only difference.

Now, there is more to the A7s than big pixels. It's been demonstrated that the A7s uses some kind of technology similar to Aptina's DRPix technology. DR Pix is dynamic range enhancement technology, operating along similar lines as MagicLantern's dual exposure (only better, more efficiently, and in the hardware). It doesn't kick in until after ISO 1600, however once it does kick in, it enhances dynamic range at higher ISO, which is why the A7s performs like a friggin powerhouse at ISO 400k. If the A7r II does not use this DR Pix like technology, then it is doubtful it would perform as well at high ISO. That said, we would once again be comparing apples and oranges...because there would be technology differences, rather than the same technology.

 

[3kramd5]

Because fill factor does not apply with BSI, big pixels or small pixels, it doesn't really matter. Everything scales, so your gathering the same total light, any given unit-area of sensor has the same photon flux, and dynamic range is the same on a normalized basis. There may be some slight real-world nuances that result in very small differences, but in general...there would be little reason to choose bigger pixels over small if you had two cameras to pick from and that was the only difference.

Now, there is more to the A7s than big pixels. It's been demonstrated that the A7s uses some kind of technology similar to Aptina's DRPix technology. DR Pix is dynamic range enhancement technology, operating along similar lines as MagicLantern's dual exposure (only better, more efficiently, and in the hardware). It doesn't kick in until after ISO 1600, however once it does kick in, it enhances dynamic range at higher ISO, which is why the A7s performs like a friggin powerhouse at ISO 400k. If the A7r II does not use this DR Pix like technology, then it is doubtful it would perform as well at high ISO. That said, we would once again be comparing apples and oranges...because there would be technology differences, rather than the same technology.

Interesting. Thanks for your thoughts!

 

[benperrin]

Now, there is more to the A7s than big pixels. It's been demonstrated that the A7s uses some kind of technology similar to Aptina's DRPix technology. DR Pix is dynamic range enhancement technology, operating along similar lines as MagicLantern's dual exposure (only better, more efficiently, and in the hardware). It doesn't kick in until after ISO 1600, however once it does kick in, it enhances dynamic range at higher ISO, which is why the A7s performs like a friggin powerhouse at ISO 400k. If the A7r II does not use this DR Pix like technology, then it is doubtful it would perform as well at high ISO. That said, we would once again be comparing apples and oranges...because there would be technology differences, rather than the same technology.

Hi Jon,

Have you decided to go with a Sony camera for your next camera or are you getting something else? I would love to hear your thoughts from an astro photography perspective. I haven't seen any high ISO examples from the a7rII which I would need to see before purchasing one. My 5dsr is still on pre-order with no release date given so I might consider the a7rII but at this stage the 5dsr is slightly ahead on paper for my needs.

Cheers,
Ben

 

[Dylan777]

1. a7rII with high MP and slower fps
2. a7sII lower MP, higher fps and higher ISO(more like 1Dx)

Forgot to mention. It WILL NOT kill DSLR

 

[3kramd5]

I haven't seen any high ISO examples from the a7rII which I would need to see before purchasing one.

(Supposed) A7RII pics, heavily downsized.

Either it's way way way pre-production (EXIF date = january 1), it's got screwy firmware, or it's fake. Either way, nothing to make any judgments from.

 

[benperrin]

I haven't seen any high ISO examples from the a7rII which I would need to see before purchasing one.

(Supposed) A7RII pics, heavily downsized.

Either it's way way way pre-production (EXIF date = january 1), it's got screwy firmware, or it's fake. Either way, nothing to make any judgments from.

Yeah, as you say still not much to go on. I'm sure it'll perform well at high ISO I'd just like to see it for myself before making any rash decisions.

 

[jrista]

Now, there is more to the A7s than big pixels. It's been demonstrated that the A7s uses some kind of technology similar to Aptina's DRPix technology. DR Pix is dynamic range enhancement technology, operating along similar lines as MagicLantern's dual exposure (only better, more efficiently, and in the hardware). It doesn't kick in until after ISO 1600, however once it does kick in, it enhances dynamic range at higher ISO, which is why the A7s performs like a friggin powerhouse at ISO 400k. If the A7r II does not use this DR Pix like technology, then it is doubtful it would perform as well at high ISO. That said, we would once again be comparing apples and oranges...because there would be technology differences, rather than the same technology.

Hi Jon,

Have you decided to go with a Sony camera for your next camera or are you getting something else? I would love to hear your thoughts from an astro photography perspective. I haven't seen any high ISO examples from the a7rII which I would need to see before purchasing one. My 5dsr is still on pre-order with no release date given so I might consider the a7rII but at this stage the 5dsr is slightly ahead on paper for my needs.

Cheers,
Ben

I am definitely getting at least one Sony camera. I would already have the A6000, however the rumors keep cropping up about the A6100 or A7000, and the specs for the AF on the newer model sounds amazing, so I've been holding out. If the A6100/7000 is pushed out too much longer, I'll probably just get a used A6000.

I am also pretty certain that an A7r II is going to be my landscape camera. I am not sure that I would spend $3200 on it...not because I don't think it's worth it, but because I simply have too many things I need to buy, and my money can only go so far. I am hoping it will drop to $2800 or around there soon enough, and I'll probably pick one up.

There are a number of reasons I'm so interested in Sony. I met a guy, Mike Malik, on an astro forum. He does DSLR and mirrorless astro modifications, so he's inside cameras from several brands all the time. He started modding Canon DSLRs, and was fairly impressed with the 6D. However, once he got his hands on a Sony (A7s I believe it was), he said on several occasions how blown away he was by the quality of them. The design, the internal layout and construction...he stated it was quite superior to how Canon builds their DSLRs. Not all the materials were ideal in some of the external cases, like the mount, but internally, apparently Sony cameras are EXTREMELY well built.

On top of that, once A7s vs. 6D comparisons started to show up, the differences were STARK. The A7s trounces the 6D, hands down, no contest, game over. With 50% more, even double, the integration time in 6D images, compared to short exposures at quite high ISO on the A7s , the A7s produces far cleaner results, with deeper exposure, than the 6D can deliver. Results at ISO 2000 and 3200 from the A7s look at least as good as ISO 100 in your average daytime photography. It's just mind boggling how good the A7s is for astro.

Now, the A7 and A7 II...they don't seem to be delivering quite the same IQ. They are good, but only marginally better than the 6D. We think the primary difference is that the A7s uses that dynamic range enhancement technology at high ISO (something similar to Aptina's DRPix technology), whereas the A7/A7 II do not. The A7r also seems to do quite well, and it certainly delivers the resolution, but even on a normalized basis, it's still not quite as good as the A7s for astro. At the moment, I think the A7s is hands down the best ILC on the market for astro, especially when astro modded. Cleanest, deepest results you can get without going to full CCD (and even then, CCD cameras require more integration time to produce similarly clean results....the CCD advantages are thermal regulation and cooling, and being mono most of the time, so they support LRGB and narrow band filtration.)



For the record, I have not seen Sony's lossy compression exhibit as a problem for astro very often. In very rare cases, you might find some compression artifacts around stars, but for the most part, the data once integrated appears very clean.

If you are interested in a Sony camera for astro, I would skip the A7/A7 II. Those will deliver marginally better quality than the 6D...and the 6D tends to be a much better buy. The A7s is really where it's at for astro on the Sony side. Phenomenal quality. We just recently had a discussion about using drizzle integration to increase the resolution of A7s images. The big pixels are great for high resolution imaging (bigger, longer scopes), but undersample with shorter scopes. Shorter scopes are much easier to image with, though. With drizzling, you can gain 50% or more resolution if you get enough subs, and process correctly. So you can get both the really clean results, as well as good resolution, by pairing the A7s with drizzle integration.

 

[jrista]

Here are a couple of examples:

http://www.cloudynights.com/uploads/monthly_06_2015/post-205769-0-45448000-1435261812.jpg
http://www.cloudynights.com/uploads/monthly_06_2015/post-205769-0-36074400-1435261831.jpg

The top is the A7s, bottom is 6D. The A7s is 42 minutes of exposure at ISO 1600. The 6D is 60 minutes of exposure at ISO 1600. The 6D subs were gathered under darker skies, by about a factor of 1.6x, which is 1 2/3rds stops BETTER than the skies the A7s subs were gathered under. The 6D was modded with a "full spectrum" mod, which totally opens the sensor up to as much light as possible by entirely removing all the filters. The A7s was only a partial mod, which removed the primary IR cutoff filter, but left all the other filters in place, so there was still some filtering occurring.

This scenario gave EVERY advantage possible to the 6D. The 6D has longer integration time by 43% (half a stop). The 6D was a full spectrum mod so it had no filtering of any kind, whereas with default filtering it doesn't even get 20% of the primary Hydrogen emission, Ha (hydrogen alpha) while the A7s still had it's first LPF filter in place which filteres out some Ha. The 6D was used under darker skies (nearly a pristine dark site, 21.2 mag/sq"), which reduces extra unwanted light signal, which reduces photon shot noise, whereas the A7s was used under skies that had a moderate amount of light pollution (20.0 mag/sq").

This all gives the 6D at least a two stop, if not a two and a half stop, advantage...and DESPITE all those advantages over the A7s...the A7s still produced a cleaner image, with more stars, with better colors, less background noise, and a deeper exposure.

Don't know about anyone else, but that is pretty telling to me. The technology Sony is packing into their cameras, whether all the external factors are perfect or not (lot of people on this forum bring up Sony's plastic mount as a major issue, for example), is truly phenomenal stuff.

 

[StudentOfLight]

Here are a couple of examples:

http://www.cloudynights.com/uploads/monthly_06_2015/post-205769-0-45448000-1435261812.jpg
http://www.cloudynights.com/uploads/monthly_06_2015/post-205769-0-36074400-1435261831.jpg

The top is the A7s, bottom is 6D. The A7s is 42 minutes of exposure at ISO 1600. The 6D is 60 minutes of exposure at ISO 1600. The 6D subs were gathered under darker skies, by about a factor of 1.6x, which is 1 2/3rds stops BETTER than the skies the A7s subs were gathered under. The 6D was modded with a "full spectrum" mod, which totally opens the sensor up to as much light as possible by entirely removing all the filters. The A7s was only a partial mod, which removed the primary IR cutoff filter, but left all the other filters in place, so there was still some filtering occurring.

This scenario gave EVERY advantage possible to the 6D. The 6D has longer integration time by 43% (half a stop). The 6D was a full spectrum mod so it had no filtering of any kind, whereas with default filtering it doesn't even get 20% of the primary Hydrogen emission, Ha (hydrogen alpha) while the A7s still had it's first LPF filter in place which filteres out some Ha. The 6D was used under darker skies (nearly a pristine dark site, 21.2 mag/sq"), which reduces extra unwanted light signal, which reduces photon shot noise, whereas the A7s was used under skies that had a moderate amount of light pollution (20.0 mag/sq").

This all gives the 6D at least a two stop, if not a two and a half stop, advantage...and DESPITE all those advantages over the A7s...the A7s still produced a cleaner image, with more stars, with better colors, less background noise, and a deeper exposure.

Don't know about anyone else, but that is pretty telling to me. The technology Sony is packing into their cameras, whether all the external factors are perfect or not (lot of people on this forum bring up Sony's plastic mount as a major issue, for example), is truly phenomenal stuff.

Couple of questions:
1) Why do the two images have different magnification?
2) What do the histograms of the integrated files look like?