Does CRI matter for scans of Black & White negatives?

koraks said:

The second plot I think comes with some caveats. It seems to represent the combined response across all three RGB channels with a particular Kolari filter fitted on the lens. This filter likely has a non-linear response to compensate for certain peaks & valleys that result from crosstalk also/in particular in the case of a modified (IR filter removed) camera. So I'm not sure how representative it is in the context of this thread.

Click to expand...

In fact, the second picture shows transmission curves of those camera's original filter. I use it because I didn't find a single curve about new Sony camera's UV-IR, or seriously, NIR cut filter. Canon hasn't released any information about its CMOS products so it's impossible to know those sensors' feature. The reason I show these transmission curves is that some high CRI light sources emanate some waves beyond 700nm, and that's why we should consider UV-IR cut filter's influence. E.g. halogen lamp and COB LED I used for my slide projector are able to give off some infrared. I show its datasheet below but sorry no English version. (BTW, my machine projects all my slides in a neutral, natural and beautiful way. But it shouldn't be discussed in this thread)

koraks said:

Note finally that the kind of curves shown here can only be constructed on the basis of measurements to which in-camera processing is already applied to normalize the signals. We cannot really know what kind of math (potentially fairly complex, even) is applied to the raw sensor data as this is manufacturer-proprietary information and occurs inside the black box of the onboard image processor. I suppose that if a datasheet of the actual sensor is available (which is probably most often not the case for consumer/prosumer cameras), it might be possible to at least crudely reverse engineer this black-box behavior to a limited extent.

Click to expand...

As for this, I quote the first picture from Viework's datasheet, which claims that "The sensitivity data may not match the measurement on the finished product necessarily because it is measured based on the wafer". So I think it can represent IMX661 sensor's original data.

koraks said:

Again, for B&W acquisition it's not very relevant either way. Things get more interesting when digitizing color film. In that case, we enter a world of still conflicting views with reasonable arguments on either side of the divide (narrow-bandwidth vs. full-spectral width), but AFAIK the 'state of the art' in professional film scanning (e.g. for motion picture use) has for a long time been narrow-bandwidth, separate R, G, B captures because this gives the cleanest channel separation and thus is capable of extracting the most color information from the dye image.

Click to expand...

Yeah, film industry has a much more serious and mature process for analog and digital. Both BMD and ARRI use RGB LED in their digitizing systems. Factories and labs can even choose prism camera to get a sharp and clear spectral distribution before. But ARRI and BMD turn to field scan system instead of line scan. I don't really know if they can fully overcome color crosstalk. Analogue process is not easy to learn and practice.

AFAIK, optical printers also use interference filter to emit pure narrowband light. I guess maybe one reason is that to some extent, traditional print films are a kind of narrowband material and use RGB can simulate print films' color somehow.

Anyway, I do plan to make a Status M LED system to take Epson V850's place for digitizing 35mm film and making positives.

MsLing said:

In fact, the second picture shows transmission curves of those camera's original filter.

Click to expand...

Okay, I stand corrected - but to my defense, it takes a reversed image search to find the probable source of the image you showed: https://kolarivision.com/uv-filter-effectiveness-article/
It really helps to include a link for contextual information esp. if it's not self-explanatory - note that there's no caption inside the image itself. My confusion stemmed from the inclusion of the same or a similar plot here: https://eu.kolarivision.com/products/kolari-vision-uv-ir-cut-hot-mirror-pro-2 In this case the plot is also poorly (not) captioned, but the fact that it's a product page, implies it's a plot of the camera with the filter mounted.

MsLing said:

I quote the first picture from Viework's datasheet

Click to expand...

This must be the product leaflet you're referring to: https://www.b2s.kr/data/bbsData/17478120641.pdf?ckattempt=1
The quoted text is on the last page.

MsLing said:

So I think it can represent IMX661 sensor's original data.

Click to expand...

Yes, exactly the problem. You're never working with data from a naked sensor. It's always processed.

MsLing said:

I don't really know if they can fully overcome color crosstalk.

Click to expand...

Undoubtedly that's not entirely possible, no. But it's relatively easy to take a multi-spectral light source and use the camera sensor as if it's monochrome for consecutive captures of the various channels.

MsLing said:

Anyway, I do plan to make a Status M LED system to take Epson V850's place for digitizing 35mm film and making positives.

Click to expand...

I'd be interested in hearing about your results with this. In a similar vein, I hope you'll also make a thread about the projector solution you mentioned before. The spec sheet of that light source looks very promising indeed; do you have a link to the product itself or at least a product name/model number?

photonaro said:

Do I understand you correctly - are you saying that color temperature or CRI of the light source should not have an impact on the contrast of an image with DSLR copies?

Click to expand...

In the context of copying a B&W image, I think you would have more variability by changing brands of cameras than seeing a difference (using a single brand/model of camera) under different color balance or CRI of illumination. But my opinion is based on subjective assessment of images, while Koraks and MsLing are presenting objective measurement curves.
Can you subjectively sense the difference between 62mph and 64mph, (if not comparing your car to another car...if both are going 62 or both going 64? you notice only differentials in speed, not absolute speed; a speedometer presents objective data so you know a 2mph change)

koraks said:

Okay, I stand corrected - but to my defense, it takes a reversed image search to find the probable source of the image you showed: https://kolarivision.com/uv-filter-effectiveness-article/
It really helps to include a link for contextual information esp. if it's not self-explanatory - note that there's no caption inside the image itself. My confusion stemmed from the inclusion of the same or a similar plot here: https://eu.kolarivision.com/products/kolari-vision-uv-ir-cut-hot-mirror-pro-2 In this case the plot is also poorly (not) captioned, but the fact that it's a product page, implies it's a plot of the camera with the filter mounted

Click to expand...

Ah, my bad. I will add a link next time.

koraks said:

This must be the product leaflet you're referring to: https://www.b2s.kr/data/bbsData/17478120641.pdf?ckattempt=1
The quoted text is on the last page.

Click to expand...

Yes, and Vieworks writes that sentence in every datasheet

koraks said:

I'd be interested in hearing about your results with this. In a similar vein, I hope you'll also make a thread about the projector solution you mentioned before. The spec sheet of that light source looks very promising indeed; do you have a link to the product itself or at least a product name/model number?

Click to expand...

The 100w COB I used is YSX-100TW/W45G14-Z3C. I will post a thread and share something about modifying projector. If the application doesn't need such a high power density, Seoul Semiconductor's Sunlike are also excellent products. I use a 5000k module for my room's lighting. But maybe I should do some simulations next time before designing in EDA , diffuser acts embarrassingly. And I think I still need some time to design a scanning system.

MsLing said:

The 100w COB I used is YSX-100TW/W45G14-Z3C. I will post a thread and share something about modifying projector. If the application doesn't need such a high power density, Seoul Semiconductor's Sunlike are also excellent products.

Click to expand...

Thanks, those are very interesting! Will look into this for sure!

koraks said:

YMMV, but it never seemed to be that simple to me. Still doesn't, really!

Click to expand...

Compared with color editing. It's all relative, my friend, as much as it is subjective