Leica Konost: a new FF digital RF body for M?


This is pretty interesting:

Some young Cornell post grads are going for it!

"At the heart of the Konost Camera is a 20.0mp 35mm Full Frame CMOS image sensor from CMOSIS. The CMV20000 is a high resolution, high dynamic range, and high speed sensor allowing us to achieve spectacular detailed images. With 16 LVDS channels capable of running at 480 Mbps, the CMV20000 can also record videos at 30fps at full resolution, giving the Konost Camera extended capabilities besides still photography (hint*). CMOSIS has also been excellent in providing documentation, reference designs, and customer support for our application."

"The Konost Prototype uses the Zedboard development kit to drive the massive image sensor. With an onboard Xilinx Zynq All-Programmable SoC/FPGA, we were able develop our own data and image algorithms, serial controls, protocols, and memory allocations, giving us full control and features of the CMV20000 image sensor. The Konost Prototype also uses the 7″ Zed Touch Display providing live preview and user interface to the sensor. This low cost approach provides further flexibility for customization and future developments."

"All features of the Konost Prototype are custom designed. From the image sensor board, to the lens mount, and even the final enclosure, all were designed to provide maximum flexibility and usability in testing and development. The Konost Prototype is a working full frame camera with 30fps live preview, complete sensor control, and raw image data acquisition. Although designed around the M-Mount lens type, the Konost Prototype can also take all F-Mount lenses via an adapter. If you’re interested in using our prototype for your own development purposes or even professional studio work at a low cost, feel free to contact us at our contact page."
There is no reason why a group of engineers cannot pull this off, but the details of the implementation are what is important.

I will be watching!

And I tried talking my Nephew into going to Cornell into Computer Engineering.



I looked up the sensor. Monochrome and Color available. The dynamic range is 66dB, 5dB less than the M8 sensor. Full well-capacity is 15K, 1/4th of that of the M8.

The real problem is going to be the microlens array, unless it is offset like the Leica sensors- there will be a lot of vignetting from wide-angle lenses and fast normal lenses.

Aren't these the folks who make the M240 sensor? Can you compare the specs to that CMOS? :)


Bakersfield, CA
Real Name

I looked up the sensor. Monochrome and Color available. The dynamic range is 66dB, 5dB less than the M8 sensor. Full well-capacity is 15K, 1/4th of that of the M8.

The real problem is going to be the microlens array, unless it is offset like the Leica sensors- there will be a lot of vignetting from wide-angle lenses and fast normal lenses.

Probably why they mention F mount lenses, which would have no such problem due to the longer register distance.
The data sheet is not published for the M240 sensor. The Dynamic range is supposed to be 76dB according to one press release. Numbers for Signal to Noise and uniformity- I can't find any. The saturation count about 40K. Honestly- without a Data Sheet, impossible to know. Most of the other 6um family seems to have a saturation count close to what the CMV20000. I'm guessing the M240 sensor is double-sampling to average out noise and increase the saturation count. "I would"...

It looks like they are building a camera around the development/evaluation kit, makes sense.

BUT- this is a 12-bit sensor, 12-bit raw files. The Dark Current to Saturation count ratio IS 12-bits.

This will be a great replacement for an RD-1. Expect vignetting with the Full-frame version. I believe the APS-C version if priced right, will be a great entry level- like an RD-2 that people wanted. Note they show an IR filter in use over the lens, not sure how they will handle the thin sensor stack. The M240 sensor has more IR bleed than does the M9.

A "Hot Mirror" filter over the lens is more powerful than the M8 IR filter, the latter filters out 95% of IR.

Hot mirror filter with full-spectrum EP2.


Which is about the same as my "regular" EPM1.

They are using a sort of in chip programmable clipping voltage level up to two times during one exposure (see end of page 20 in the linked datasheet below). Similar to the RED cameras. This boosts the DR from 66db (11 stops) to 90dB (15 stops). The designed generic glass cover is of 0.7mm. The buyer can specify the glass type...

Thanks for the Link- I downloaded the data sheet, in my collection now. The data sheet states that IR absorbing glass should be placed in the optical path for a color sensor. Kodak changed the cover glass for me 20+ years ago, so I know it can be done. A run of 50 detectors was made, so there is some minimum order. Some sensors used Taped cover glass, makes it easier. I'd rather have the Monochrome IR version of the camera. I wonder how much the development camera will be.

Page 36 gives angular response, which may be an issue with wide-angle and fast normal lenses.

Personally, I'd rather see a Still camera using a CCD:


With a 14-bit A/D development kit available. 71dB linear dynamic range.
Last edited by a moderator:


San Francisco
I just read all the links associated with Konost's website, very exciting. I'm not up to speed on all the technical aspects that others have posted to this thread but are the developement kits for larger companies that might want to use Konost's technology to develop other products (i.e. lenses, accessories, apps, etc.)?

I always wondered why there weren't more companies that make digital rangefinder bodies, particularly the larger companies that did make anologue rangefinders. I understand that rangfinders in general are a niche market, but why doesn't Canon or Nikon put out a digital rangefinder rather than trying to compete in the mirrorless market. I'd love to see Canon re-release their Canon 7 rangefinder in a digital version.
Well Nikon I fans will be excited. The sensor is 32mm x 24mm.

They should do this camera in S-Mount. Then they can use a shaft encoder easily, geared into the mount.
I would love to see another camera to replace the RD-1. I don't think this one is it. 12-bit pixels are not sufficient for the resolution of the sensor. They need to use a better sensor to match the optics. If the camera comes in at under well $2,000- I can see people buying it. I would - just to play with the software development kit. Higher price than that, a used M9 is better on all specifications for still photography.

My advice for the project- find a better sensor, or keep it really cheap.
Last edited by a moderator:

I wish they would fix the specification sheet for the sensor image size. They state the sensor image size measures 36mmx24mm (3:2 ratio) and give the sensor resolution as "5120(v) x 3840(h)" (4:3). They got the "V"ertical and "H"orizontal wrong, it is correct on the CMOSIS site. Looking at the viewfinder of the camera, the sensor is not flipped in the body.

If this were April 1st, I would think it was a joke.

I would like to see a Digital RD-2, and I would like to see a Digital RF with an open-source firmware that I can play with. I just cannot take this camera seriously until they examine the specifications of the sensor that they plan on using. It's not suited for an M-Mount.
Last edited by a moderator:
If I seem harsh on Konost, it's from being an Engineer in Optical Sciences for over 35 years. I've hired Cornell Engineering students, and worked for a Cornell graduate.

Points off for not checking basic math.


Last edited by a moderator:
I would like to know what the thickness of the body is- it has to be ~28mm Plus the thickness of the sensor, board that it sits on, with the LCD screen behind it. To put a number on it: a standard accessory shoe on a Leica is ~17.5mm. A Canon P film camera is 31.6mm (I used calipers) for the top plate. The mount extension and back extends a little farther, ~5mm. The M9 is 37mm deep, so a little more than double the hot shoe depth.

If they want to keep the body thin, it looks like they need to position the mount in front of the camera, like a Nikon RF and Contax. I'm not sure the flush mount as shown leaves enough room for the sensor, LCD, and electronics. They could leave the LCD off the back and use a Hybrid viewfinder for all menu settings, review, etc. That would allow the camera to be thinner compared with the M9. The thickness of the sensor from the rear of the pins to the image plane is 6.8mm, not counting the socket that it must fit in. I'd ditch the LCD in back of the camera and keep it thin, use menu selection on the hybrid screen like the M60.

As shown, this camera is not well-thought out.
Last edited by a moderator:

Latest threads

Top Bottom