I have some interesting findings regarding backlight sources that could be of use to some out here, so I wanted to share these to “give back” for the many things I learned on this forum, and also to get some inputs. I recently created a new table-top scanning setup for 35 mm, after some previous adventures with an up-side-down enlarger, since this was more convenient:
This enabled me to try multiple light sources: A Solux bulb in a metal mount (which gets rather hot), a recently acquired more convenient (cheap, compact, bright, and not getting hot) Rollei Lumen Square (~35 EUR), which can be controlled at various color temperatures through a phone app, and my iPad. Logic has it that the 99+ CRI daylight bulb should be better than a cheap LED panel, correct? So I did some tests that I wanted to share. The old Nikon PS-6 slide and negative holder has a glass diffuser mounted very close to the negative/slide but not touching it. This is as it standard comes.
- Take (RAW) pics of glass diffuser in PS-6 without negative/slide with fixed aperture (f8)
- Decrease shutter speed until peaking occurs with Rawdigger (ETTR)
- Back-off 2/3 stop and use that picture in LR with NLP2.3 profile
- Use white balance picker in LR to see how “white” the light is
- Drop exposure to look for vignetting and color cast
Bare Solux bulb at distance as in pictures above
Rollei Lumen with it’s extra diffuser at about 20 mm from PS-6
iPad Pro 9.7" (2016) directly against PS-6
Bare Solux with extra opal glass diffuser (see pic above) about 5 cm from PS-6
Rawdigger histograms, left to right: Bare Solux, Rollei at 4100K, iPad, Solux + extra opal glass
(I do not know what the bottom G2 (2nd green channel) in Rawdigger represents)
Horizontal is EV, vertical is as far as I understand the amount of pixels with that EV level. Scales for all channels per light are the same.
I noticed that for the bare Solux bulb, the R, G, and B histograms did not have the same shape. This leads to color-casts in a scan since the R, G, and B channels do not overlap after whitepoint setting with the LR color picker:
(center = white, but color shifts towards blue in the corners. The slight vignetting is most likely due to the lens, as I found out, since it’s also there and similar when the light evenness is varied.) The non-equal histograms for the various channels in the Rawdigger plot cause the color cast. Eventually I traced this down to the Solux requiring more diffusion (see below).
The cheap Rollei has impressively similar histograms for R, G, and B, and this provides nice overlap in LR after a color picker WB in the center:
This white light should give accurate colors, correct?
The iPad has sharper histograms, which would translate in “better color separation” and seems more “ideal” after reading in various places here about proper RGB light sources(?), but is less bright and takes 3 seconds shutter speed per shot (without film). The consistent shape of the histograms results in good overlap and no color casts in LR:
(blue in extreme corners is peaking of black)
I was puzzled about the blue color casts of the Solux towards the corners so I placed an additional opal diffuser in between the Solux and the PS-6. This cleaned up the RAW histogram and made the individual channels similar in shape (right pic in the Rawdigger plot above), and resulted also in removing most of the color cast in LR:
The individual histograms still do not overlap as good as for the Rollei though.
Lessons learned from this exercise:
- Rollei impresses and is convenient and bright, as well as enables different color temperatures which some claim is better for slides vs. B&W vs. negatives.
- Solux takes some doing to get homogeneous and neutral but is continuous spectrum with CRI99+. Opinions seem to vary as what is best: High CRI continuous, or well defined and separated RGB. With the Rollei, and my lens, sensor, and other in-betweens, I obtain smooth neutral grey across the entire area, so I should get neutral colors as well (no?), provided I do the same ritual with the Rollei on an unexposed section of film?
- iPad impresses but its light is too weak for my setup.
Overall, I conclude that the Rollei will do just fine and the Solux is not worth the hassle of working with an extremely hot light and less convenient setup, although I still worry that LED’s are not continuous spectrum (but neither is my camera’s sensor).