GFX100, a 15 fold increase of value compared to an XF IQ350 (per Bernard)

[BernardLanguillier]

I’m sure that true. Nothing wrong with that. Amateurs, not fond of the word as sometimes it can seem insulting and I don’t mean that at all, have the luxury of deciding what they want to say and how they want to say it and then can choose their tools accordingly. A pleasant position to be in.

Absolutely. I am an amateur myself.

The only meaning I associate with that word is that of not generating a majority of income from photography.

Cheers,
Bernard

[Rob C]

Absolutely. I am an amateur myself.

The only meaning I associate with that word is that of not generating a majority of income from photography.

Cheers,
Bernard

That's an interesting philosophy.

It has already destroyed the professional stock photography market. Thank you, shamateurs united, for that.

Rob

[SrMi]

Absolutely. I am an amateur myself.

The only meaning I associate with that word is that of not generating a majority of income from photography.

Cheers,
Bernard

noun: amateur;
1. a person who engages in a pursuit, on an unpaid rather than a professional basis.
synonyms:   nonprofessional, nonspecialist, layman, layperson
2. a person who is incompetent or inept at a particular activity.
synonyms:   bungler, blunderer, incompetent, bumbler;

The term amateur is ambiguous. Maybe we should use terms nonprofessional or blunderer, depending on what we want to convey :-).

[BernardLanguillier]

That's an interesting philosophy.

It has already destroyed the professional stock photography market. Thank you, shamateurs united, for that.

Well, as far as I am concerned I don’t make any income from photography and have always refused to give away images for free except to personal friends.

I would have said blunderer speaking of myself but I didn’t want to risk offending other amateurs.

Cheers,
Bernard

[eronald]

Well, as far as I am concerned I don’t make any income from photography and have always refused to give away images for free except to personal friends.

Cheers,
Bernard

I think Rob needs a beer
The cellphone destroyed the photo market, as people assume that they can take pictures so they have no value.

Edmund

[faberryman]

The cellphone destroyed the photo market, as people assume that they can take pictures so they have no value.

I was at the doctor's office today. There were two or three dozen magazines laying around. The amount of photos greatly exceeded the amount of text. I'd wager most were not taken with an iPhone, and that none of them were taken by photographers for free.

[mondeo]

It is my belief - founded on my 40 year old education in IC design, that chips are priced by surface. (The root cause of this is that defects follow a Poisson statistic).

Edmund

Well then your 40 yr education will remind you that for each chip process step improvement / reduction the cost of masks to make the chip tends to double in price. Compared to a 50Mp chip you need photosites that are half the size and a process geometry that is smaller, otherwise all the amplifiers will take up proportionally more space. Also the likely yield to get a good enough sensor reduces (you need half as many dead pixels or dead amplifiers).

This all assumes that the number of layers the chip has hasn't also increased. More layers in processing also equals more cost even if the area hasnt changed.

All of this is standard chip manufacturing economics and not Fuji (or Sony) pulling a fast one.

[eronald]

I was at the doctor's office today. There were two or three dozen magazines laying around. The amount of photos greatly exceeded the amount of text. I'd wager most were not taken with an iPhone, and that none of them were taken by photographers for free.

https://www.tvtechnology.com/opinions/how-euronews-nbc-approaches-iphone-journalism

e.

[eronald]

Well then your 40 yr education will remind you that for each chip process step improvement / reduction the cost of masks to make the chip tends to double in price. Compared to a 50Mp chip you need photosites that are half the size and a process geometry that is smaller, otherwise all the amplifiers will take up proportionally more space. Also the likely yield to get a good enough sensor reduces (you need half as many dead pixels or dead amplifiers).

This all assumes that the number of layers the chip has hasn't also increased. More layers in processing also equals more cost even if the area hasnt changed.

All of this is standard chip manufacturing economics and not Fuji (or Sony) pulling a fast one.

Mondeo,

Go ahead and educate a dinosaur

The number of interest here is the relative cost factor of a 44x33 chip compared to a 36x24, when the same process and cell architecture is employed. I think this is determined by the process yield /die size curve.

I don't think one-time setup costs like maskset are really relevant because I would expect them to be well under $200K.

My expectation is that Sony ran the numbers for something like 10K chips - with 20 chips per 300mm wafer - if that's what they're using these days, and a 1500sqmm die size if one factors in 2mm on each side of the chip for scribing and pads.

However, I suspect that Sony is rather good at optimising its sensor production mix using multichip wafers, so the production of the larger chip might occur seamlessly at the same time as whatever else they are making with that sensor-tuned process.

I think that's enough made up numbers for one day. As you are well informed, I am sure you can provide accurate yield figures for the process Sony is using for this product, and we should be able to figure out the incremental chip cost better than by my guesses.

Edmund

[Dan Wells]

Going back a bit...

In unit sales, APS-C vastly exceeds everything larger put together - even Fuji's upper-end APS-C is a relatively small piece of the market. APS-C (including Micro 43) is somewhere between 3/4 and 4/5 of ILC sales (something like  7.5-8 million units/year).  The vast majority of these are inexpensive "Best Buy cameras",  for which no additional lens is ever purchased (they're used with the 1-2 lenses they came with).

Of these, somewhere between 500,000 and 750,000 are Fujis with a much higher average value per unit than most other APS-C cameras - and many of those Fuji shooters buy lenses.  A few more are other higher-end APS-C cameras, although a serial number reporting site estimates total Nikon D500 sales over 3+ years as only 110,000+, compared to millions per model of the D3xxx and D5xxx series. I'd be surprised if the Canon 7D mkII or the Sony A6500 were selling in a completely different order of magnitude from the 30,000-40,000 D500s per year. Most $1000+ APS-C is Fuji...

Full frame is probably somewhere around 2 million cameras per year (this figure plus or minus 20% keeps showing up different places), at much higher unit prices (probably not 4x the price per unit, but maybe not far off, especially if you include lenses). Many or most FF shooters own several lenses for their system, and some own lenses worth much more than the camera body.

I'd be shocked if medium format is even 100,000 cameras/year - it may now be close to that, with the Fujis selling much more than the body/back systems ever did. Average selling price is very high.

BUT

Non-Fuji APS-C almost certainly has an average price/camera kit (body plus lenses owned) well under $800 ($600?)
Fuji APS-C is probably $1500-$2000 (body and a couple of lenses)?

Full-frame is at least $2000, probably significantly higher - depends on lens sales - how many of those cameras share a bag with an expensive lens or two?

Is medium format $10,000 or so per kit? Most of the market is under or around that, but there's Phase and the H system getting towards $50,000/system.

Total revenue from full-frame is probably approaching total revenue from APS-C  (especially if you count Fuji X-T, X-H and X-Pro as competing with full-frame, not with $400 cameras). Total profit from full-frame may even be higher than APS-C, because the margins are higher.

Medium format is much smaller, but not insignificant - 1/20 the sales of full-frame, but at 5x the price? 1/4 the revenue???

[hogloff]

I was at the doctor's office today. There were two or three dozen magazines laying around. The amount of photos greatly exceeded the amount of text. I'd wager most were not taken with an iPhone, and that none of them were taken by photographers for free.

Magazines...another thing going the way of the dodo bird.

[hogloff]

Going back a bit...

In unit sales, APS-C vastly exceeds everything larger put together - even Fuji's upper-end APS-C is a relatively small piece of the market. APS-C (including Micro 43) is somewhere between 3/4 and 4/5 of ILC sales (something like  7.5-8 million units/year).  The vast majority of these are inexpensive "Best Buy cameras",  for which no additional lens is ever purchased (they're used with the 1-2 lenses they came with).

Of these, somewhere between 500,000 and 750,000 are Fujis with a much higher average value per unit than most other APS-C cameras - and many of those Fuji shooters buy lenses.  A few more are other higher-end APS-C cameras, although a serial number reporting site estimates total Nikon D500 sales over 3+ years as only 110,000+, compared to millions per model of the D3xxx and D5xxx series. I'd be surprised if the Canon 7D mkII or the Sony A6500 were selling in a completely different order of magnitude from the 30,000-40,000 D500s per year. Most $1000+ APS-C is Fuji...

Full frame is probably somewhere around 2 million cameras per year (this figure plus or minus 20% keeps showing up different places), at much higher unit prices (probably not 4x the price per unit, but maybe not far off, especially if you include lenses). Many or most FF shooters own several lenses for their system, and some own lenses worth much more than the camera body.

I'd be shocked if medium format is even 100,000 cameras/year - it may now be close to that, with the Fujis selling much more than the body/back systems ever did. Average selling price is very high.

BUT

Non-Fuji APS-C almost certainly has an average price/camera kit (body plus lenses owned) well under $800 ($600?)
Fuji APS-C is probably $1500-$2000 (body and a couple of lenses)?

Full-frame is at least $2000, probably significantly higher - depends on lens sales - how many of those cameras share a bag with an expensive lens or two?

Is medium format $10,000 or so per kit? Most of the market is under or around that, but there's Phase and the H system getting towards $50,000/system.

Total revenue from full-frame is probably approaching total revenue from APS-C  (especially if you count Fuji X-T, X-H and X-Pro as competing with full-frame, not with $400 cameras). Total profit from full-frame may even be higher than APS-C, because the margins are higher.

Medium format is much smaller, but not insignificant - 1/20 the sales of full-frame, but at 5x the price? 1/4 the revenue???

Where are you getting these figures from?

[eronald]

Where are you getting these figures from?

maybe
http://www.cipa.jp/stats/report_e.html

[mcbroomf]

Mondeo,

I don't think one-time setup costs like maskset are really relevant because I would expect them to be well under $200K.

Edmund

I'd be interested where that number came from.  I was working in a 90nm factory until I retired and a full set of reticles cost in the range of $1m.  21 layers I think for a 6 level metal process.  I'm not sure how many layers of metal sensors have, a few less so remove 2 reticles/metal layer.  Now double that for a 2 stitch, etc.  The same costs have to be born for the processor if it's rev'd to add more oomph, single set of reticles though.  I would think these costs would be passed on by the semiconductor company in the initial wave of processors, quickly dropping in price as Intel does for leading CPUs.

[eronald]

I'd be interested where that number came from.  I was working in a 90nm factory until I retired and a full set of reticles cost in the range of $1m.  21 layers I think for a 6 level metal process.  I'm not sure how many layers of metal sensors have, a few less so remove 2 reticles/metal layer.  Now double that for a 2 stitch, etc.  The same costs have to be born for the processor if it's rev'd to add more oomph, single set of reticles though.  I would think these costs would be passed on by the semiconductor company in the initial wave of processors, quickly dropping in price as Intel does for leading CPUs.

Mike -

 I'm probably closer to wrong, and you're probably closer to right. As you've noticed I'm not very careful, precise or well informed.
 Like you say, one would expect a sensor design to need fewer metal layers than complex logic.
 In any case one would end up with a figure of $1M by your calculation (provided the design doesn't need to be iterated), and also wafer costs etc which are not trivial because there are fairly few dies per wafer, and the yield isn't necessarily high, so you only get a few working chips per wafer.

 But still, after all of this, if you agree that they can do a fairly fast design cycle from cell libraries, then I'd guess the whole story basically has  (company internal) economics that are closer to an ASIC design than to a processor.  Sony basically has an internal sensor design ASIC shop in-house speccing and then quickly creating chips that are suitable for cellphones, automotive, security, machine vision and ... medium format cameras. That's the business model of their Sony Semiconductor Solutions BU, although they do take the risk that a design won't sell.

 Do you agree my number of $1000 per chip to the customer still makes sense?  I would expect them to have an initial order of something like 3000 from Fuji, and they have been making this chip for some time for other applications, eg Phase One aerial.

Edmund

[chez]

maybe
http://www.cipa.jp/stats/report_e.html

I don't see the numbers broken out by APS-C, full frame and medium format. Just Interchangable lens cameras and built in lens cameras.

[mcbroomf]

Mike -

 I'm probably closer to wrong, and you're probably closer to right. As you've noticed I'm not very careful, precise or well informed.
 Like you say, one would expect a sensor design to need fewer metal layers than complex logic.
 In any case one would end up with a figure of $1M by your calculation (provided the design doesn't need to be iterated), and also wafer costs etc which are not trivial because there are fairly few dies per wafer, and the yield isn't necessarily high, so you only get a few working chips per wafer.

 But still, after all of this, if you agree that they can do a fairly fast design cycle from cell libraries, then I'd guess the whole story basically has  (company internal) economics that are closer to an ASIC design than to a processor.  Sony basically has an internal sensor design ASIC shop in-house speccing and then quickly creating chips that are suitable for cellphones, automotive, security, machine vision and ... medium format cameras. That's the business model of their Sony Semiconductor Solutions BU, although they do take the risk that a design won't sell.

 Do you agree my number of $1000 per chip to the customer still makes sense?  I would expect them to have an initial order of something like 3000 from Fuji, and they have been making this chip for some time for other applications, eg Phase One aerial.

Edmund

I definitely agree that the cell libraries can be re-used, and probably shrunk as necessary, although the smaller features would have to be made in silicon and tested before new libraries are released for the design teams to use.  It would not surprise me if Sony have leapfrogging design teams like Intel uses.

I really don't recall much about silicon costs now, and I worked in a 200mm factory not 300mm. I think the masks are more likely to be closer to $1.5M than one, and with modest sales, say 3000 then this would add $500 to the die cost assuming Sony wish to recover the cost in the 1st round of sales.  However even if you were not taking this into account and were using $1k (even though you originally said less than $1k) a cost of $1500/die is still not unreasonable I'd think to absorb into the cost of camera.  Personally I don't think the $10k price is either incomprehensible nor overpriced.  I remember paying $8k for all 3 models of the Canon 1Ds bodies I owned, and with inflation I'd bet they would be over $10k now. 

If you want bleeding edge then be prepared to pay bleeding edge prices, if not then wait.  One thing we all know is that as tech drives forward prices drop, either by discounts or the next gen being better and getting more for less.  Personally I hope they bring out a cheaper, smaller non-IBIS body in a couple of years.

[eronald]

Mike,

I think this 100MP design was done for remote sensing/ mapping applications, where I’d guess cameras are used in clusters of maybe 9 or so. Such a camera has been sold by Phase aerial/defense for some time now, and so I think the first round of sales is already done.

As for the price of the camera, I don’t think it is exagerated per se - especially when compared to Phase One and Hassy pricing in the past. I just think it is a *bad* price because Fuji have one genuine window IMHO chance to establish the system as a studio reference, and therefore should go for market share.

The Canon 1Ds was a special case, it was unique, and was used to test the waters. My feeling is Canon never wanted to sell that many, and indeed they soon came out with the low priced 5D series which was very strongly marketed.

Edmund

I definitely agree that the cell libraries can be re-used, and probably shrunk as necessary, although the smaller features would have to be made in silicon and tested before new libraries are released for the design teams to use.  It would not surprise me if Sony have leapfrogging design teams like Intel uses.

I really don't recall much about silicon costs now, and I worked in a 200mm factory not 300mm. I think the masks are more likely to be closer to $1.5M than one, and with modest sales, say 3000 then this would add $500 to the die cost assuming Sony wish to recover the cost in the 1st round of sales.  However even if you were not taking this into account and were using $1k (even though you originally said less than $1k) a cost of $1500/die is still not unreasonable I'd think to absorb into the cost of camera.  Personally I don't think the $10k price is either incomprehensible nor overpriced.  I remember paying $8k for all 3 models of the Canon 1Ds bodies I owned, and with inflation I'd bet they would be over $10k now. 

If you want bleeding edge then be prepared to pay bleeding edge prices, if not then wait.  One thing we all know is that as tech drives forward prices drop, either by discounts or the next gen being better and getting more for less.  Personally I hope they bring out a cheaper, smaller non-IBIS body in a couple of years.

[BernardLanguillier]

Japan, the country where packages get delivered on Sundays. Ah you are really far from your native Europe, Bernard.

It looks like I won’t have to tap into their weekend delivery potential in the end.

Cheers,
Bernard

[mcbroomf]

Mike,

I think this 100MP design was done for remote sensing/ mapping applications, where I’d guess cameras are used in clusters of maybe 9 or so. Such a camera has been sold by Phase aerial/defense for some time now, and so I think the first round of sales is already done.

As for the price of the camera, I don’t think it is exagerated per se - especially when compared to Phase One and Hassy pricing in the past. I just think it is a *bad* price because Fuji have one genuine window IMHO chance to establish the system as a studio reference, and therefore should go for market share.

The Canon 1Ds was a special case, it was unique, and was used to test the waters. My feeling is Canon never wanted to sell that many, and indeed they soon came out with the low priced 5D series which was very strongly marketed.

Edmund

I've never worked in marketing, although been in many meetings with them, not for leading edge tech mind you.  The Fuji is cheap for MF compared to P1/Hassy, I think we agree, so I imagine Fuji chose their pricepoint carefully, knowing that they are unlikely to get mainstream 35mm FF users to switch given the lenses (physically large but smaller range).  A bad pricing structure may be a matter of opinion, or something we'll just have to wait for proof, either way.  One thing I do know is that if they chose a low price they can't increase it, while if their MSRP is too high and limits sales, they can drop it.  Lenses are already $500 off across the board.