The mp "war"... which side are you on?

[ErikKaffehr]

Hi,

The first part is very simple. Developing a new sensor takes something like three years. We see announcements more often, but it is because development of sensor generations is interleaved. When sensor A is delivered the next generation B has been in pipeline for 1.5 years.

So, Phase One now has the IQ3-100 MP sensor. It has a pixel pitch that is probably pretty normal with 0.18 micron design rules. Now, let's assume that Phase One wants to have a 56 MP sensor with 6 micron pixels. So they task Sony with the development of that sensor. So in three years they will have that new sensor, but they have to finance that development in full.

So, making two different set of sensors with different technologies will have at least twice the costs. That would be just fine if Phase could sell twice the number of low res backs at the same price. But, that is not very probable.

Now, someone may argue that 6 microns is an ideal pixel size, but that may not hold very long when competition introduces 4.5 microns on larger sensors.

I have very little doubt that 4.5 microns will deliver on well designed platforms and with competent users. Would that not be the case, Phase One would not be able to sell their high end backs at the present prices.

Getting back to the optimal pixel size, I would refer to this article: http://isl.stanford.edu/~abbas/group/papers_and_pub/pixelsize.pdf

They define image quality as DeltaE at the pixel level. So a perfect reproduction would yield DE=0.

From the enclosed picture you can see that minimum DE is arrived at around 6.5 microns for 0.35 micron design rules and around 4.5 microns for 0.18 micron design rules. Now, this article is a bit dated, pixel designs have improved with recent designs. Would you have serious interest in  the issues at hand, I would suggest reading the quoted paper.

Best regards
Erik

I'm not sure I understand this...  what is doubling the development costs? ...additionally, what is an "optimal" size pixel and who defines it? ...Do you care to explain?

As far as I know, Leica and Sinar are the only ones that have set a "red line" with the minimum size of pixel they use (at 6μm) and seem to keep it despite the sensor size... I would trust their judgement better than other makers, since they are the most experienced ones in imaging among all makers with proven and continuous contribution to quality as well as being famous to be satisfied of nothing but the best with respect to competition...

[alan_b]

Chris,

Why did you pass on the 50MP CMOS backs for the Sony?

Totally, Theodoros.  If you regularly work with a view/tech camera.  The shift of going from dark images on a groundglass to Live View that allows 100% zoom is absolutely huge.  It has been the single biggest improvement to my overall on set workflow in the last 20 years.  Seriously.  I'm playing with the IQ3 100 a little more today, testing my Blad glass.  Having decent live view is the only reason I would ever consider this back.  I just wish zooming in and navigating a 100% image was as quick as the Sony.

-CB

[vjbelle]

Erik..... thanks for the informative response.

Victor

[Chris Barrett]

Chris,

Why did you pass on the 50MP CMOS backs for the Sony?

I didn't really see any advantage over the A7r2.  Also, if I'm going to shoot MFDB, I want a big sensor to take advantage of wide glass.  Actually, you've got me kind of wondering what the 23 HR would look like on that 50.  Though I've heard it has issues with shifts.

[bbrantley]

I just wish zooming in and navigating a 100% image was as quick as the Sony.

You're the first person to call this out, but I noticed it, too.  It's also about half as slow as zooming the IQ150.  Workable, but unfortunate that we had to go backwards.

[Wayne Fox]

It's also about half as slow as zooming the IQ150.

twice as many pixels to deal with, maybe the current back doesn’t have room to beef up the processing power to handle the load.

[Theodoros]

Hi,

The first part is very simple. Developing a new sensor takes something like three years. We see announcements more often, but it is because development of sensor generations is interleaved. When sensor A is delivered the next generation B has been in pipeline for 1.5 years.

So, Phase One now has the IQ3-100 MP sensor. It has a pixel pitch that is probably pretty normal with 0.18 micron design rules. Now, let's assume that Phase One wants to have a 56 MP sensor with 6 micron pixels. So they task Sony with the development of that sensor. So in three years they will have that new sensor, but they have to finance that development in full.

So, making two different set of sensors with different technologies will have at least twice the costs. That would be just fine if Phase could sell twice the number of low res backs at the same price. But, that is not very probable.

Now, someone may argue that 6 microns is an ideal pixel size, but that may not hold very long when competition introduces 4.5 microns on larger sensors.

I have very little doubt that 4.5 microns will deliver on well designed platforms and with competent users. Would that not be the case, Phase One would not be able to sell their high end backs at the present prices.

Getting back to the optimal pixel size, I would refer to this article: http://isl.stanford.edu/~abbas/group/papers_and_pub/pixelsize.pdf

They define image quality as DeltaE at the pixel level. So a perfect reproduction would yield DE=0.

From the enclosed picture you can see that minimum DE is arrived at around 6.5 microns for 0.35 micron design rules and around 4.5 microns for 0.18 micron design rules. Now, this article is a bit dated, pixel designs have improved with recent designs. Would you have serious interest in  the issues at hand, I would suggest reading the quoted paper.

Best regards
Erik

For the first part, (your reasoning about how cost doubles) I would say that it is totally out of the logic science rules as it is based on assumptions and conclusions that are ficticious...  It will take a book to reply on your reasoning, so I'll just say this... P1 (or Hasselblad or Pentax or anybody) doesn't go to Sony and say "Hi Guys..... we want you to make an 100mp sensor for us...", what happens is the other way around... Sony informs makers that there is a new sensor coming and checks if the sensor interests some or all the makers and if it does, asks for an estimation of the quantities that each maker will need as to price it according to the demand....  For a maker to ask for a specific sensor, it requires to approach a maker who makes exclusive sensors under requirement... Then the sensor is exclusive to the one that orders it and the same applies to the rights of it... (Leica does that on their later Cmos sensors).

Now, as far as the article is considered (which you've posted again a few months ago), it assumes that the light is directed towards the pixel directly (not in an angle), so it is rather generous in the pixel size it suggests...

As I said before, I would trust much better Leica engineers about what is "optimum" pixel size than any forum "Einstein" or than Sony... Reason for the "forum Einsteins" is obvious, reason for trusting Leica more than Sony is simply because Sony's marketing has a reputation for "quantity is all that people understand" approach, while Leica addresses their products (again traditionally) to photo-graphers and their needs.

[vjbelle]

You're the first person to call this out, but I noticed it, too.  It's also about half as slow as zooming the IQ150.  Workable, but unfortunate that we had to go backwards.

First I've head of this also......  My Leaf is fairly fast - I've never had an issue with zoom speed.  I'm also assuming that the FPS rate is not affected by the larger files. 

Victor

[pixjohn]

I would rather have less MB and better quality. What I really want is a 50mb back that works well with my Cambo Tech camera and Schneider lenses. I am still shooting with a Aptus 75 and a Iq140 because no real upgrade works without having to get new glass and start over again. The image quality has been very good since day 1 when I bought my back new. I also don't need to have such large files to work with and store.

[Theodoros]

I didn't really see any advantage over the A7r2.  Also, if I'm going to shoot MFDB, I want a big sensor to take advantage of wide glass.  Actually, you've got me kind of wondering what the 23 HR would look like on that 50.  Though I've heard it has issues with shifts.

Same problem here (having a view camera compatible without having to convert the rear standard of it with both MFDB & mirrorless)... I ended up (heavily) converting a Rollei x-act2 which will end up compatible with every lens from 45mm mount distance FF DSLRs and up to any LF lens ever made... It will also accept (with portrait - landscape rotation) any mirrorless or MFDB on it by simply detaching one and attaching the other.... Additionally, it will have focal plane shutter and full interface control of the MF lenses that have electronic aperture or electronic leaf shutter (so that it will be compatible with multishot backs too)... Contact me if you are interested... I have a state of the art machining lab backing up the project and the end product (all my design) will be in production by the end of the year... (end product will have no similarity with Rollei x-act2 at all - the choice of the Rollei is only to test the new standards).

[ErikKaffehr]

Hi Theodoros,

Sorry to say, but your postings have a pretty low signal to noise ratio…

Best regards
Erik

For the first part, (your reasoning about how cost doubles) I would say that it is totally out of the logic science rules as it is based on assumptions and conclusions that are ficticious...  It will take a book to reply on your reasoning, so I'll just say this... P1 (or Hasselblad or Pentax or anybody) doesn't go to Sony and say "Hi Guys..... we want you to make an 100mp sensor for us...", what happens is the other way around... Sony informs makers that there is a new sensor coming and checks if the sensor interests some or all the makers and if it does, asks for an estimation of the quantities that each maker will need as to price it according to the demand....  For a maker to ask for a specific sensor, it requires to approach a maker who makes exclusive sensors under requirement... Then the sensor is exclusive to the one that orders it and the same applies to the rights of it... (Leica does that on their later Cmos sensors).

Now, as far as the article is considered (which you've posted again a few months ago), it assumes that the light is directed towards the pixel directly (not in an angle), so it is rather generous in the pixel size it suggests...

As I said before, I would trust much better Leica engineers about what is "optimum" pixel size than any forum "Einstein" or than Sony... Reason for the "forum Einsteins" is obvious, reason for trusting Leica more than Sony is simply because Sony's marketing has a reputation for "quantity is all that people understand" approach, while Leica addresses their products (again traditionally) to photo-graphers and their needs.

[ErikKaffehr]

Hi,

I guess that the prospects of having a CMOS MFD back designed for near symmetric wide angles is pretty bleak. The IQ3-100MP sensor is in all probability designed for MFD exclusively. But I guess that Sony (and Phase One if they were involved in the design) put emphasis on usability with DSLRs and excellent DR rather than resistance to cross-talk.

The sensor used on the Leica M240 is designed to work with symmetric lenses and some early info from Leica indicated that the CMOS sensor in the Leica S (type 007) is a similar design. But, as far as I know that design is proprietary to Leica and I have not seen any information that it would be used on a general purpose MFD back.

There has been some speculations that it may be used on Sinar camera designs, but no such camera has been materialised yet, as far as I know.

Best regards
Erik

Unfortunately I don't think many share this view, and even if they do it's not reflected in their buying behavior. Medium format is dependent on providing higher pixel counts than whatever is provided in the smaller formats. It's what medium format does. At some point it must level out, but I think we're not there yet.

Huge file sizes can be a mess when you don't need them but a better solution than low pixel counts is sensor+ type of things.

My big thing is that I miss the situation where you could choose a digital back that was designed to handle symmetrical wide angle lenses. The Kodak I'm using has both light shielded pixels (minimizes crosstalk) and lacks microlenses (minimizes artifacts such as ripple). I think the pixel size today could be made smaller than the 6um and still retain the same properties, possibly 100 MP sensor would work just fine. However one needs to face the fact that tech cams are now dinosaurs, it's not likely that a sensor will ever be designed for that again, especially since light shields and removing microlenses will surely lead to tradeoffs in ISO performance and possibly DR and aliasing.

What will happen is that CCD products will be soon all discontinued (due to poor sales), there will be a continued increase in megapixels, and entry level products are based on 44x33mm crops of the full-frame sensor. We won't see larger sensors than 645 fullframe.

I still have a small hope that BSI/stack sensors as a side effect can make tech wides work well again but it may be too late when that arrives.

In the long term I want very high pixel counts as I think the optics should be a limiting factor and I want my images 100% aliasing free. The lowest pixel count I'd like to have today is about 60MP. Very low counts is an aliasing mess.

[Theodoros]

Hi Theodoros,

Sorry to say, but your postings have a pretty low signal to noise ratio…

Best regards
Erik

I'm surprised you say this Erik... isn't there enough reasoning backing up my posts? Just because you think different on size of pixels than Leica decided, it doesn't mean that that you are right and they are wrong..... there may be a (small  ) chance that they have find something you... missed! 

[jsiva]

First I've head of this also......  My Leaf is fairly fast - I've never had an issue with zoom speed.  I'm also assuming that the FPS rate is not affected by the larger files. 

Victor

I find the IQ180 fast enough, so not sure why the 100MP would be "slow".  Has anyone compared the 80MP vs. 100MP?

[vjbelle]

Maybe there should be clarification regarding Chris's post.  From my point of view he was describing how the 100MP back performed while zooming in at 100% for LIVE VIEW..... not for viewing/playing files at 100%.  My IQ180 in live view at 100% is lacking to the point that I found it pretty much unusable - not only because of the slow FPS rate but also for the high noise level of the LCD image.  My Leaf is in a different world much like the Nikon/Sony although maybe not quite as fast but almost there.  If the 100MP back is half the speed of my Leaf I'm OK with it..... at least so far. 

Victor

[jsiva]

Maybe there should be clarification regarding Chris's post.  From my point of view he was describing how the 100MP back performed while zooming in at 100% for LIVE VIEW..... not for viewing/playing files at 100%.  My IQ180 in live view at 100% is lacking to the point that I found it pretty much unusable - not only because of the slow FPS rate but also for the high noise level of the LCD image.  My Leaf is in a different world much like the Nikon/Sony although maybe not quite as fast but almost there.  If the 100MP back is half the speed of my Leaf I'm OK with it..... at least so far. 

Victor

Ah, got it.  In LV, the Sony is quite fast.  I was referring the IQ180 in review.

[bbrantley]

The live view FPS appeared roughly identical to me (IQ350 v IQ3 100) and plenty fine for focusing. 

What was about twice as slow was the time between the double-tap and a 100% view being acquired on screen.  Something like 2.5 to 3 seconds was common, although it varied a bit for reasons I couldn't discern in the short time I had.  It's as if the back just needs more time to go schedule the right rows and columns for readout or something.  I wonder if they might be able to tune it down the road in firmware.

Unfortunately for my case, I do a ton of moving around and zooming to verify focus, so I that extra lag is annoying to contemplate having to put up with.

[Chris Barrett]

Exactly.  The way that I am accustomed to judging focus on the Sony is zooming into the liveview and then moving that around, which provides realtime response.  Doing the same on the back is very sluggish.  I suppose it would do that much faster off the capture, but then you can't focus 'live'.

[Ken R]

Oh gawd yes.  People will go on and on about signal to noise ratios and barely be able to say more than 'I like that...' about the actual work.  Man, I love me some gear, but at the end of the day I just want to make beautiful photographs and have people 'get' them.

Well, it is a tech oriented forum but for the love of all that is....photography... it would be great if the tech talks were contextualized with the art and craft of making images for a purpose (even fun). Not just testing for testing's sake and complaining. I mean it's not like most of the people who are more engineering oriented in this forum can do anything about their findings (unless they get a job with the camera/back MFG) other than deciding their own purchases and criticizing other's. They come to some great conclusions and almost can reverse engineer anything but man, there needs to be some balance.

Everything has to be put in the context of what it was designed for.

I mean reading some of the posts I cannot imagine any of these people shooting positive film (E-6 or Kodachrome) ever. Can you imagine Yunli shooting slide film? with such little dynamic range.

[synn]

Everytime soneone here complains about DR, Bruce Percy creates anothing kickass image on Velvia.