What are Hasselblad up to with their 16 bit color space claims?

[Chris Dodkin]

I received an invitation from Hasselblad's marketing company to attend a live webinar on their X1D camera system last week, 'X1D-Great Camera and Great Buy', hosted by forum regular Eric Peterson, Hasselblad Field Sales Specialist.

The webinar was an overview of the camera system, with a live video stream, and chat based technical Q&A.

You can review a recording of the webinar here: Hasselblad Webinar Recording

As part of the webinar, Eric reviewed the benefits of the Sony sensor in the X1D. (Starting around 8 mins in to the recording.)

Then, at approx 10:27 Eric states:

This camera system is also a camera system that captures and moves the data into a 16 bit color space, so that gives us over 65 thousand variations of color as opposed to 16 thousand for a 14 bit color space that some of the other camera systems are using.

This immediately struck me as an odd statement to make, as the Sony sensor has a fixed 14 bit digital output - there is no 16 bit data available from the sensor.

I immediately asked about this via the online chat.

The tech Q&A session answered may of the chat questions, but this question was not addressed on the day.

Instead I received an Email from Eric as follow-up, some days later:

Chris,

Thank you for attending the X1D Webinar.

My reference to bit depth explained that data is moved off the sensor into a 16 bit color space. This doesn’t add any information but ensures that we don’t lose anything in all transformations/calculations performed on the sensor data. It was not my intention to mislead anyone.

Thanks,

Eric Peterson

Field Sales Specialist | Hasselblad Inc.

It was great to receive an answer to my question, however the justification given does not match what was said during the webinar.

What was said during the Webinar, specifically pointed out that the X1D benefitted from 65,000 shades of color (per R/G/B channel) vs 16,000 on 14 bit systems.

This is at best misleading, the digital output from the Sony sensor is only 14 bit, so would not have 65,000 shades of color. It would, in fact, have the same number of shades as other cameras using the same base sensor model, from Pentax and Fujifilm. Both of these companies claim 14 bit color.

The Email justification appears to be based on the concept that later manipulation of that 14 bits of data, is now calculated into a 16 bit container, ensuring Hasselblad 'don’t lose anything'.

I can appreciate that during the live event, the presenter could push a marketing agenda - but to follow up and maintain that the 16 bit packaging gives them sort sort of color IQ advantage seems to me to be a triumph of marketing over science.

Erik has already done some work on this here: https://www.dpreview.com/forums/post/59002429

What does everyone think?

--
Your time is limited, so don't waste it arguing about camera features - go out and capture memories

 

[JimKasson]

Instead I received an Email from Eric as follow-up, some days later:

Chris,

Thank you for attending the X1D Webinar.

My reference to bit depth explained that data is moved off the sensor into a 16 bit color space.

This seems to imply that the data is no longer stored in raw form. But, looking at the raw files, the data is clearly mosaiced. Is is possible that X1D raw cookery extends to using data from adjoining pixels? What would be the point of that? I doubt if that is what's going on and that talk of changing the color space is at best imprecise.

I suspect that instead the processing that takes place in-camera is related to distortion or vignetting correction, or something similar. Since many cameras don't bother to byte-pack 14-bit data, there is little cost to leaving it 16-bits.

This doesn’t add any information but ensures that we don’t lose anything in all transformations/calculations performed on the sensor data.

Was does everyone think?

The key questions are:

What processing takes place to get from 14-it sensor data to 16-bit raw data?

What is the correlation of the last 2 bits of the 16-bit data and the image projected upon the sensor. If the last two bits are essentially noise, the answer is none.

Jim

 

[Jones0610]

I also watched that webinar and thought it was "interesting" but not especially informative. Struck me as being overly fluffy and short on specific differentiators between the X1D and competitive products. The speaker clearly was not an Engineer and was simply parroting talking points.

I also picked up on the 16 bit color comment but since Hasselblad is using the same sensor that is in other MF cameras I didn't assume that they are capturing and processing more colors than anyone else using similar designs. I did, however, get the impression that the speaker was trying to fast talk misinformation hoping no one would notice.

There are quite a few VERY negative reviews out there on the X1D, some posted by people who typically have little or nothing negative to say about anything they are reviewing. It seems that the major selling point for the X1D is it's body design... and it is certainly an attractive looking camera. But in terms of features, stability, usability and so on, it certainly leaves me wanting more.

I am not at all a fanboy but based on the impartial reviews I've read here and elsewhere it seems to me that the GFX 50S offers the most bang for the buck right now. I was hoping that webinar would address the camera's long startup time and buggy firmware issues... instead they completely ignored them like no one who's paying $13,000 USD for a camera wouldn't do sufficient research to know about these issues before buying.

 

[jacketpotato]

What is the correlation of the last 2 bits of the 16-bit data and the image projected upon the sensor. If the last two bits are essentially noise, the answer is none.

Jim

 

[Paul]

this 14 vs 16 bit discussion is so old.....

there is a lot of stuff and fluff to read in google searches, i think its just weird that hasselblad even goes there.....

if the sensor puts out 14 bit, that is all the info there is....you can wrap it up afterwards in all you want....and honestly i don't think there is a difference between 14 and 16 bit on screen or on paper...there are so many other factors to IQ and tonality and hasselblad is doing a great job with both, so i think its just silly to throw numbers around to make a product sound better....

but hasselblad knows their clients and that a lot of people are or will buy into the X system because it is hasselblad and because it offers 16 bit....at least leica makes their own glass.....

i have no problem with the X1D at all, but i do believe that the GFX is all around the better camera and system, at a better price....and i completely understand if one wants to buy the hasselblad because that is what they want to shoot with and they will get great results.....

the negative reports about the system are i think more a reaction to hasselblads own claims and or unrealistic positive reports from people who think they have to protect a brand.....no matter how pretty a camera is, if there are crashes and lost images when the camera comes out, those are issues that cant be brushed off IMO......afaik all the main issues have long been resolved and i am sure it now is a dependable camera.....

thanks for posting a link to the video, i signed up but couldn't watch.....a year after first playing with this camera, i still haven't been able to get a good answer on how the wifi feature (i really like in the H system) actually works on the X....was hoping there is something in there about it.....

 

[jacketpotato]

"This camera system is also a camera system that captures and moves the data into a 16 bit color space, so that gives us over 65 thousand variations of color as opposed to 16 thousand for a 14 bit color space that some of the other camera systems are using."

Sony sensor has a fixed 14 bit digital output - there is no 16 bit data available from the sensor.

What was said during the Webinar, specifically pointed out that the X1D benefitted from 65,000 shades of color (per R/G/B channel) vs 16,000 on 14 bit systems.

This is at best misleading, the digital output from the Sony sensor is only 14 bit, so would not have 65,000 shades of color. It would, in fact, have the same number of shades as other cameras using the same base sensor model, from Pentax and Fujifilm. Both of these companies claim 14 bit color.

What does everyone think?

 

[NAwlins Contrarian]

This camera system is also a camera system that captures and moves the data into a 16 bit color space, so that gives us over 65 thousand variations of color as opposed to 16 thousand for a 14 bit color space that some of the other camera systems are using.

One hesitates to use descriptions like 'fraud', but IMO this is either misleading or indicative that some of Hasselblad's other claims--of outputting raw files--are at least somewhat misleading.

* I can't imagine that there are too many people using image processing software that would process 14-bit image files with only 14-bit precision, instead of 16-bit. So the truncation or rounding errors inherent to mathematical operations where the precision is no greater than the least significant bit would seem like a non-issue. Sending 14 bits of data in a 16-bit container does not seem capable of increasing quality.

* That is, unless the camera is not simply passing along the 14-bit output of its analog-to-digital converters. As someone else said, if the camera is 'baking in' some sort of data processing to the 'raw' files, then there would be an advantage to having two extra bits less significant than the A-to-D converter's actual data, to minimize the impact of rounding or truncation on whatever processing the camera performed. But then of course the resulting camera output would not be true raw files. Yes, I realize that 'baking in' would be nothing new; e.g., Sony has (rightly, IMO) taken a lot of flak for putting out 'raw' files that were unavoidably processed in-camera to reduce noise. I'd think the sort of people looking at a camera in the X1D's class would strongly favor the camera merely tagging any suggested processing, and letting the user make the final decisions on what to do at the raw conversion stage, using software options.