LargeSense LS45 now available for purchase

[Oren Grad]

Looks like it's finally become a reality.

Monochrome version only at this point. 6.7 MP, 4.7 x 5.5 inch capture area. $26,000.

http://largesense.com/index.php/products/4x5-large-format-digital-back-ls45

[Kiwi7475]

Amazing news. Was wandering how to burn this $26k pocket change myself.

/s

[Mark Sawyer]

6.7 megapixels? Wow, that's almost as much as an obsolete cell phone camera...

[Oren Grad]

Plenty for your Petzval...

[Michael R]

I’m trying to understand the technical benefit. Is it that with such large pixels you get a very high native sensitivity/very low noise?

I’m also confused by how the colour version will work. How much “effective” resolution will you get if you only start with 6.7MP monochrome?

Not that I could afford this thing anyway even if it turns out to be awesome, but anyway.

[Mark Sawyer]

Plenty for your Petzval...

Only for internet posts. Petzvals are among the sharpest lenses in the image center.

[Oren Grad]

Only for internet posts. Petzvals are among the sharpest lenses in the image center.

Fair point!

I think the 8x10 (9x11, actually) is more interesting at 26MP and with less need for enlargement, but of course the cost will be even further out of reach.

[Jim C.]

Just curious, the example pics on the linked URL shows that a Schneider Kreuznach G-Claron 9/305
has swirly bokeh like a petzval ?

[Robert Tilden]

Anyone know what the sensor specs are? Is it appropriated from some other piece of equipment? X-Ray machine perhaps? Given the size and resolution could it be a smaller sensor connected with a fiber array to the focal plane?

[Oren Grad]

Anyone know what the sensor specs are? Is it appropriated from some other piece of equipment? X-Ray machine perhaps? Given the size and resolution could it be a smaller sensor connected with a fiber array to the focal plane?

From the linked page:

Specifications:

4.7 x 5.5 inch high sensitivity, large pixel digital sensor
6.7 Mega Pixels - 6,720,000 pixels
50-micron pixel size (these are very large)
Live view for easy focusing
WiFi for remote viewing and control
Video mode with an optional 1920 x 1080. Up to 28 fps. Video output options TIFFs, DNGs, CinemaDNG, JPEGs, h.264
14-bit AD
No CFA, no AA, and no micro lenses. This allows large movements without affecting edge image quality.
Output to DNG, TIFF, and JPEG
Media: SD card slot, USB 3.1 external drive, internal 1,500 GB SSD
Tethered operation, by CAT5 cable or WiFi
The sensor plane is around 22mm behind what would be the film plan for the Graflok model. This may affect SLRs, but assumed not mirrorless cameras.
Dust - The sensor is large and can pick up dust. It is easy to remove the bellows and clean the sensor. Also, an FFC frame can be taken before a photo session, which works very well in removing most dust. The FFC frame is incorporated into the camera software.
All specifications subject to change

[wsit]

I have been following their development on the 4x5 and 8x10 sensor.

There are a lot of bells and whistles like deep neural net software to add more resolution to the capture. High numbers of network connection for users, video capture, etc. As a professional sw engineer in ML and data intensive discipline, every one of these features require a lot of r&d resource and domain experts to deliver high quality product. I see that this company needs a hw engineer, embedded system engineer, UI software engineer, data processing engineer, etc. none of them comes cheap.

I have my double that this is a well funded and well resourced company. If that’s the case, there are a lot of effort spent on non essential features to the large format folks.

25k to spend on 6mp very large and ‘very clean’ pixels for 4x5. I see that the lower end hasselblad and p1 back being superior in both hardware and software offering.

I am their ideal customer based but I don’t see unique upside of this product over the blad and p1..

I hope they find an industrial application to keep funding this effort so I might eventually get a high MP 4x5 and 8x10 digital sensor system that doesn’t require a truck and dolly to haul around.

[Oren Grad]

I have my double that this is a well funded and well resourced company....

From the "About" page:

The smallest cameras are made by the biggest companies, and the biggest ones by the very smallest. Our research indicates that the market for large format digital cameras is tiny, so LargeSense needs to be tiny. If Canon and Nikon could be thought of as elephants, Phase One would be a kitten and LargeSense would be a ladybug (Coccinellidae).

Our secret sauce is building and selling a specialty camera where the market is only a few cameras per year. It is not something that even a small medium format company could do profitably. Keeping costs low while building a revolutionary new camera is required.

http://largesense.com/index.php/about12

[wsit]

From the "About" page:

The smallest cameras are made by the biggest companies, and the biggest ones by the very smallest. Our research indicates that the market for large format digital cameras is tiny, so LargeSense needs to be tiny. If Canon and Nikon could be thought of as elephants, Phase One would be a kitten and LargeSense would be a ladybug (Coccinellidae).

Our secret sauce is building and selling a specialty camera where the market is only a few cameras per year. It is not something that even a small medium format company could do profitably. Keeping costs low while building a revolutionary new camera is required.

http://largesense.com/index.php/about12

Right. Exactly my point. There can’t be that much money in this niche. Says they have 100 4x5 customer a year ( grossly over estimate imho), it’s only 2.5mil in sales. I have my further double you can find 100 new customer every year for this product. You can’t hire that many engineering talents with this sort of budget. I hope they can laser focus on what matters to this client base. Perhaps this existing client bases really want video feature, a auxiliary machine learning software, multi client network connectivity, etc. if not, I don’t want to invest in a piece of hardware where the company might go out of biz in a few years, with (assumed) closed source software that unlikely would work should the biz close.

I hope this company has a long term sustainable biz model because I appreciate any new and precious resources dedicated to large format photography, and just don’t want this biz to fizzle out….

[reddesert]

I couldn't find out much about the detector technology. Is it CCD or CMOS? Or something else? The high frame rate for video suggests a CMOS-like technology to me, but I have not heard of anyone making such large-pixel large devices in CMOS. On the other hand I haven't really heard of anyone making such large devices as monolithic CCDs, either. I sort of favor "something else."

Honestly what surprises me about the cost is that it's not 5 or more times higher. Unless there is a magic trick, this is absolutely enormous for a semiconductor device. I wonder if under the hood, it's some kind of mosaiced array. Or if it is fashioned from something like a digital radiography detector minus the X-ray sensitive part. I can't see how a semiconductor foundry would be able to develop and make a few devices a year without charging a million bucks per detector; it must be part of some larger run.

I agree that neural net image processing sounds like an elaborate add-on, but it is possible, or likely, that post-processing is necessary to remove detector artifacts. Neural network, by itself, could mean any number of things. Large semiconductor detectors usually have artifacts and the number of flaws you can tolerate has a strong influence on the yield of a foundry run and hence the cost (eg "science grade" and "engineering grade" CCDs).

[Oren Grad]

A couple of articles with a few more technical and operational details:

https://ymcinema.com/2021/07/06/meet-the-worlds-largest-sensor-camera-that-shoots-raw-the-ls45-sensor-size-of-140mmx150mm/

https://ymcinema.com/2021/07/09/ls45-huge-format-camera-concept-dynamic-range-workflow-and-pricing/


Honestly what surprises me about the cost is that it's not 5 or more times higher. Unless there is a magic trick, this is absolutely enormous for a semiconductor device. I wonder if under the hood, it's some kind of mosaiced array. Or if it is fashioned from something like a digital radiography detector minus the X-ray sensitive part. I can't see how a semiconductor foundry would be able to develop and make a few devices a year without charging a million bucks per detector; it must be part of some larger run.

Good point.

< scratching head >

[Dan Fromm]

The reference to cinematography has me scratching my head.

[reddesert]

One of those YM Cinema links says it's a CMOS detector, although it's just tossed off and articles sometimes get details wrong.

There are large CMOS detectors used for medical radiology and non-destructive testing, for example (just from googling) Hamamatsu makes a 120mm sq, 2.4x2.4K pixel, 50 micron sensor: https://www.hamamatsu.com/us/en/product/optical-sensors/x-ray-sensor/x-ray-flat-panel-sensor/x-ray-flat-panel-sensors-for-radiology/index.html . And Teledyne DALSA makes a number of ~100 micron pixel sensors, and a 228x291mm, 4.6x5.8K pixel, 50 micron sensor: https://www.teledynedalsa.com/en/products/imaging/medical-x-ray-detectors/xineos-large-area/ Someone in one of the old threads suggested Dalsa as a possible source.

These X-ray sensors have an X-ray scintillator deposited directly on the surface of the CMOS detector, so it's not as simple as taking one off the assembly line, but it tells us that such a large CMOS detector can be built, and perhaps gives an idea of why it has pixels as large as 50 microns. I am sure that even with an existing detector, the engineering challenges in repurposing it for pictorial use are significant.

[Oren Grad]

A sensor originally targeted for clinical or industrial X-ray applications sounds entirely plausible. Something amusing about it, too - if silver halide film marketed for X-ray use can be repurposed for pictorial use, why not sensors?

Dan, can you find a new digital camera these days that's *not* also specified for video use?

[SergeyT]

"Digital" wins over "Analog" mostly because of convenience (or promises of elusive "savings").
In this case the back adds inconvenience to already inconvenient large format setup and also, with its low resolution sensor, undermines the benefits of using large format gear compared to other smaller formats. Let's not dive into myths about magic qualities of "fat" pixels. If that was true then we won't move past Canon 1Ds circa 2002 in the full frame "digital" world.
Do you see anything really appealing in this setup, other than the ability to project on the sensor through an "exotic" lens (and possibly brag about it) ?

[Peter De Smidt]

I'm afraid I agree with Sergey on this. I'm not sure who the target market for this is. I do wish them well, though, and hope they succeed.

[Kiwi7475]

With regular (full frame, etc) sensors having pixel sizes in the 3 to 5 um, a pixel of 50um must surely have a lot more sensitivity and be able to do much better comparably in terms of noise and low light conditions— assuming the technology they use scales in performance?

[Peter De Smidt]

Remember back in the D700 days? People would argue that we reached some kind of quality level and that increasing resolution father would lead to a loss of quality. Well, that was just wrong.
People also like to talk about fat pixel Phase backs. Well, 15 years ago, I worked in a big commercial studio with 10 'fat pixel' Phase backs. I would take a D850, or even my D600, over one of those systems in a heart beat. That quality was not very good. Moire was a huge problem, as were color shifts with use with a view camera.

People also used to talk about how CCDs where inherently superior to Cmos chips.....

[Bernice Loui]

Some rough aprox numbers,

1mm/50 microns = 20 LPM. Diffraction at f64 about 22 LPM.

50 micron sensor theoretically has lower noise than a 5 micron sensor, typical view camera images are not done in low light conditions.

Essentially, the market for a solid state imager back of this size would be extremely limited.
Only time will tell where this large area solid state imager back will end up.



Bernice

[Kiwi7475]

Not just low light but how many of us doing landscape have had to wait for a long time for the wind to stop to avoid getting tree branches/leaves motion? Add to that a sunrise or sunset with the usual lower light conditions, and you’re screwed — you just can’t wait. Some people prefer using 400 iso film for that reason. But color slides don’t come in iso over 100. I’d love to shoot at speed of 1/250 or 1/500 a lot of times to get around these issues. 50um pixels should offer a distinct capability to go to significantly higher iso’s while keeping the same noise level of their 5um counterparts.

[Oren Grad]

This article from three years ago...

https://www.diyphotography.net/the-worlds-first-single-shot-8x10-large-format-digital-camera-can-be-yours-for-a-mere-106000/

...says that the 9x11 version has a base ISO of 2100.

[bernardlanguillier]

It’s great to have these options but I cannot help but think that an IQ4-150 on an Arca is a much better option in terms of actual results. It’s even more expensive though.

[jnantz]

Looks like it's finally become a reality.

Monochrome version only at this point. 6.7 MP, 4.7 x 5.5 inch capture area. $26,000.

http://largesense.com/index.php/products/4x5-large-format-digital-back-ls45
it cost a lot less than before .
only a matter of time it will be less expensive. Sadly I sold 1 kidney for a box of TMX film and I'm going to need the other one until the chimera is big enough.
what I'd love is a XEROX 11x14 camera .. sadly they are extremely rare.

[abruzzi]

The reference to cinematography has me scratching my head.

Actually, while I don’t expect it’s something many cinematographers are asking for, movements + motion actually sounds pretty cool. I don’t know if cinematographers ever use a PC lens to do interesting stuff? Or mount it on a geared 4x5 and attach servo motors to the gears, and change perspective it real time. If I had infinite funds I’d love to try some stuff like that (in another life I was a very amateur cinematographer and editor for friends who were going to film school. I still have my Bolex somewhere.)