Recent pet project: Emulation of Aerochrome as faithfully as possible via single-shot digital means

[Fandyus]

Hello everyone, I am back after quite a long break. Life got in the way.

About a month and a half ago now, I had a short term obsession with emulating the look of the legendary Aerochrome made by Kodak. It was a much needed distraction from the stress of school graduation, that's all behind me now, thankfully.

It led me to better understanding of how the actual film worked.

For those of you who don't know, the sensitivity of Aerocheome is thus:

Red forming layer: NIR up to around 850nm + blue

Green forming layer: visible red + blue

Blue forming layer: visible green + blue

It makes sense then, that people used this film with yellow and orange filters, which resulted in the film being a one shot IRG, to this day it's the only medium to ever do this, and might be the only one to ever exist, which is a huge shame. I'd buy a camera with a bayer array that emulates the sensitivity of Aerochrome any day.

But some people have realized that you have something similar with normal bayer arrays, mainly due to the transmission characteristics of the RGB filters in the Bayer array.

Sensitivity of a conventional sensor with the hot mirror removed is thus:

red photosites: red and infrared, essentially a longpass filter

green photosites: green and infrared

blue photosites: blue and infrared

Yann Philippe, an inovator in the field of ditital multispectral photography and the creator of the Kolari IR chrome has outlined a method of channel subtraction in one of this longer videos.

https://www.youtube.com/watch?v=WxWDrbYJCsA&t=1681s

By taking a full spectrum photo with an orange filter, you leave the blue layer with only the IR signal. You can then subtract this signal from the other two remaining layers, and theoretically, you should have three channels with separate information for infrared only, red and green, just like Aerochrome.

I tested this rather extensively and I am sorry to report that it does not work 100%, at least not with my Sony made sensor. The resulting two non-IR channels are still polluted with IR even after the subtraction, but less so. Someone should try with cameras that have sensors by different manufacturers to see if transmission of other Bayer array photocells allows for this to be done better.

Regardless, it still looks rather nice. It even reproduces the beige-yellowish Aerochrome skin pretty well. Unfortunately enough, it also diminishes the color depth of the image as opposed to just a simple channel swap or Aerochrome emulation using Wavelengthpro. I will follow up on Wavelengthpro in a future post.

[Fandyus]

So turns out that I can't upload jpeg files, all you get is black frames as seen above. Admins, please do notify if you need any assistance or when this is fixed so that I can update the post with proper images.

I could also just use PNGs but that will waste your bandwidth.

[Fedia]

I'm looking forward to see your images !

[Fandyus]

22 minutes ago, Fedia said:

I'm looking forward to see your images !

Thank you! Though if you've seen my instagram you've probably seen a few of them already.

[Fandyus]

Ok, it works now. Images uploaded.

[Fedia]

Very nice ! The images are carefully graded.

 

You didn't mention the exact filter(s) you used. Orange I guess, but do you have more details ? I would like to know more about your channel mixer setting also, did you do heavy IR substraction ?

 

It's strange to see the teapots come out blue like car windows. I didn't know ceramics had IR-absorbing properties.

 

 

[Fandyus]

3 hours ago, Fedia said:

Very nice ! The images are carefully graded.

 

You didn't mention the exact filter(s) you used. Orange I guess, but do you have more details ? I would like to know more about your channel mixer setting also, did you do heavy IR substraction ?

 

It's strange to see the teapots come out blue like car windows. I didn't know ceramics had IR-absorbing properties.

 

 

Thanks a lot, I try to put a lot of effort into color grading.

I used a Rowi GO-2 orange filter. It's a budget brand so no precise transmission info is available. I did not use the channel mixer at all for this. The process I used is explained in the youtube video I linked. The precise amount of IR subtraction is unknown to me, it was made through the calculations tool in photoshop, where the blue channel was layered over the red and green channels in "subtract" mode. Resulting images were pasted as the a and b channels in Lab.

The teapots are indeed odd, no idea what caused it. I don't think it's bad color balance since everything else looks pretty neutral.

Thanks for the comment.

[Fedia]

I'll have to watch the video more attentively. The technique he uses is surprising, he doesn't really explains why he does it this way instead of in the channel mixer. I don't have much time right now, I hope to be able to do it soon. I'm surprised I don't see any pink pop out in your images.

[Fedia]

I rewatched Yann philippe's video. Did you try using the channel mixer before trying this particular technique ? I wonder how different the result would be...