https://www.imaging-resource.com/news/2020/08/18/fujifilm-film-simulations-definitive-guide
2 Likes
Well, he did a lot of work. However, it doesnāt sound too hard. It all boils down to profiling. Shoot a color target on real film, shoot it on digital, compare, make a LUT or something, done. Or am I over simplifying?
Edit: I mean, surely that is how the Fuji cameraās do this internally. Take the raw values, apply some well chosen transformation, output JPEG, done.
1 Like
If I understand it correctly, the latest Fuji cameras do rather more than applying a LUT. I believe the engineers used hardware properties of the sensor, for instance when adding grain differentially to areas of the image with different luminosities, or adjusting the hue and saturation of highlights and shadows. (This is why Fuji cannot backport some film simulations to earlier cameras that have less-advanced sensors.)
In my experience, third party āFuji-style film simulationsā are simply never as good as Fujiās OOC JPGs. And Iāve spent hours trying to reproduce the exact quality of Fuji JPGs from my RAW files with post processing, and have never managed to. I now just save myself a lot of time, effort and frustration, and use Fujiās brilliant JPGs. (After all, they designed the film stock, and have decades of experience adjusting the characteristics of their film. Read the excellent article that the OP cites to get a much fuller explanation.)
To do a bit of adjustment to the JPGs (to compensate for minor issues caused by my framing or exposure settings) I just pull the Input Color Profile earlier in the pipe to linearise the workflow, as AurƩlian advises.
1 Like
I have an X-T3 and I actually vastly prefer using the film simulation HaldCLUT pack from RawTherapee on the RAW files (in darktable). You can of course get great looks out of the in-camera presets, but the HaldCLUTs are much more accurate emulations of each stock IMO (as opposed to options like āClassic Chromeā), and you give yourself far more room for error in terms of exposure instead of aiming for a good WYSIWYG exposure in the viewfinder. The latter all-too-often leads to obliterated blacks or highlights.
I always shoot RAW + JPEG to give them a chance, but yeah, as good as the jpegs are compared to the competitors you still canāt beat a RAW file.
@black_daveth Care to put that claim to the test? Could you share an OOC JPEG and a JPEG processed from the same RAW file but with a comparable HaldCLUT applied?
2 Likes
which claim?
I can show you an OOC JPEG compared to my preferred processing of the RAW file to see what you think, but thatās going to be highly subjective and hardly a test.
There shouldnāt need to be a test to show that I can capture far more dynamic range in a RAW file as compared to a JPEG and thus have more latitude for processingā¦
And the other more objective claim is that the HaldCLUTs do a better job of emulating the film stocks they claim to than Fujiās in-camera settings. Testing this would require shooting as many of the films (some of which arenāt even available any more) as possible side-by-side with the Fuji and doing a three-way comparison.
Unfortunately I donāt really have the gear or the time available to do such a test, but I would recommend checking this out: Film Sim vs Actual Film - Fuji Acros Comparison
As you can see, the results are nothing alike, but in the cameraās defense, the optics are different and both the development and scanning process can have huge effects on the resultant film image - but I believe this only reinforces my point: if the film image is what you see in the mindās eye, the JPEG canāt reliably get you that out of the box, and if post processing is required youāre far better off having the depth of a RAW file to work with.
And that example is just in black and white, while I donāt have any similar side by side comparisons for colour, Iāve seen hundreds of my Dadās Velvia and Kodachrome slides and youād only need to look at half a dozen to know the equivalent Fujifilm simulations donāt come anywhere close. Iād actually say ditto for the HaldCLUTs as well, i.e. they do very poorly also. Those two stocks are straight up magical and put on a pedestal for a reason.
As for the article, the whole āWhy canāt I just do this in Photoshop?ā argument is a bit of a non-starter. Perhaps you canāt drag on a few sliders to emulate film, but its certainly not beyond the capabilities of a satisfactory 3D LUT (which is kind of evident in the 2D plots he uses to show how different they all areā¦). This is where the Photog bloggers probably need to stop banging the drum for Fujiās marketing department and maybe learn something from the film industry where LUTs have been used in critical roles for some 20 years, perhaps more.
Quick background on that, print film emulation LUTs werenāt created to help make peopleās youtube videos look more cinematic, they were used when digital intermediate became a thing (i.e. a movie was shot on film, scanned for VFX and colour grading, then printed back to film for distribution). So that final printing to film will inherently result in all kinds of complex colour shifts, (which is just film being film of course), and the role of the PFE LUT is to emulate this perfectly for the colourist to be graded āunderā, and then removed before its sent to the printers. HaldCLUTs for what its worth are even more detailed than the industry standard cube LUTs.
Anyway, back to the beginning Iām more than happy to compare a straight OOC JPEG to a processed RAW, but like I said thatās going to be pretty subjective for the most part. If youād like to devise some other feasible test Iām definitely keen to dive a little deeper into this, its interesting stuff.
Edit: Iāll also quickly add that the HaldCLUT workflow makes it many times easier to shoot, and get consistent results, with two different cameras as well (Fuji X-T3 and Nikon D700).
Edit2: Perhaps a āfairā way of doing it would be for archiemac to supply a JPEG+RAW (seeing as heās probably put more care into his settings and knows how to get the best result in camera) and Iāll adjust the RAW in a similar way to my liking for the sake of comparison? Just a thought, if that sounds like a fun idea 
2 Likes
ok, incoming highly subjective testā¦
so the first of each image pair are the SOOC images, taken using these settings (minus the WB shift): Portra 400 recipe
and the RAWs were processed using the Portra 400 NC 2 HaldCLUT from the RawTherapee pack with the same, camera selected white balance.
first of all Iām actually quite impressed with various aspects of some of the SOOC shots, particularly the little pixel peeper inside of me can see a massive difference in detail and sharpness at 100%ā¦ But compared to all the film photography Iāve seen, the Fuji processing is never heavy handed enough to match IMO (Iām desperate to try Classic Neg if it ever comes to my X-T3 because that seems to be a little more extreme).
its a shame I donāt have anyone around to take a photo of, because where I really see a big difference between the Fuji X Weekly recipes and the HaldCLUT approach is in the skin tones. Theyāre too often flat and bleached looking for me out of camera, but you can so easily make them as rich and warm as you like with the RAW.
what I havenāt found so easy with the Fuji files is bringing back the highlights. They feel much easier to bring back on my D700.
2 Likes
but @black_daveth, PORTRA on a FUJI camera? Madness, indeed. 
1 Like
@black_daveth Thank you for the extensive response. This was the particular claim I was referring to:
I always shoot RAW + JPEG to give them a chance, but yeah, as good as the jpegs are compared to the competitors you still canāt beat a RAW file.
Allow me to go on a small ramble here, nothing against you personally, I promise. The point with discussions like these are that they are, as you rightly say, very subjective. Thatās way I am always very wary if articles claim A is better than B without some evidence to back that up. Even for subjective matter you can do a thing called ABX testing that can support claims that a certain look is recognizable from another.
There is no actual comparison to physical film anywhere in the article @darix links. So claims like āWriting this article, I was struck again and again by how much some of the film simulations matched my memories of shoeboxes full of photo prints and box after box of color slides from my childhood.ā make no sense at all. Has Fuji really done a good reproducing their famous film looks? I have no clue, because he didnāt show me. Instead he went on an extensive tour of characterizing film looks with ātypical color thisā and ātypical contrast thatā with only his memory as comparison. That makes any argument only as good as the people who believe you on your word.
However good JPEGs look that come out of a camera, with whatever secret sauce applied to them, you need to do a proper test if you are making comparative claims. @black_daveth The website you linked does that much better, thank you for that.
1 Like
Great stuff! Now only if we could have put those in an ABX test to see if people would be able to distinguish OOC JPEGs from processed ones 
I would say there are a few discussions going on here and some are less subjective than others.
- Do Fujiās film simulations do anything that canāt be achieved later in software, whether with a LUT or otherwise: Fairly objective, but difficult to know for sure without peering inside the black box.
- Do Fujiās film simulations look the same as the films they are based on: Should be reasonably objective, but difficult to test properly for reasons already mentioned.
- Is a RAW file more malleable than a JPEG: Objective, and we know the answer is yes.
- Do you prefer Fujiās film simulations straight out of camera to any conceivable RAW edit: Completely and utterly subjective.
So when I say you canāt beat a RAW file, Iām saying that for reasons related to points 1, 2 and 3, not 4.
Just to make this bit clear, these example images above are of the subjective kind. Iām not trying to match the JPEGs here, its merely a comparison of what you get out of camera to my preferred workflow when Iām trying to go for that āfilmā feelingā¦ and to get that feeling, evoking memories of your shoebox full of photo prints is ultimately far more important than a side by side comparison of a colour chart or otherwise verifying that they way I think Portra 400 would have rendered the scene is anywhere near reality. Ultimately thatās not important.
But having said that the charts are still necessary to get there, and I honestly as anecdotal as this is, to my eye RawTherapee or whoever compiled that collection of HaldCLUTs has done their homework. Fuji can talk about their secret sophistication all they like, I think the LUTs do a better job in the real world.
If youāre interested in film emulation or ācolour scienceā as it effects the photographer in general these two demos from cinematographer Steve Yedlin are absolutely essential viewing (but more so the Display Prep Demo, its a bit more relevant here):
1 Like
True! I hadnāt thought that far ahead.
Hereās another example I have on hand to further muddy the waters:
So the first image is straight out of my X-T3 on the Eterna simulation, and the second is the final edit I arrived at using the RAW file and the Fuji 400H 2 LUT from the RawTherapee pack.
Obviously very different film stocks, but I guess my only point all along is no matter how much you like them, why limit yourself to shooting JPEG only?
2 Likes
That is a lovely cat. Iāmma download that cat for wallpaper purposes. Oh also, prefer the raw edit; thereās something about the slightly enhanced orange in the fur. <3 love that cat
1 Like
Just to be clear: I am not interested in what anyone thinks looks best. My take on the original article, however, was that the premise is that Fuji definitely tries to emulate their original physical film look. The author says as much,
āFuji has more than 85 years of color science experience, and a deep understanding of all the complexities of how film works. Theyāve drawn on this background and extremely detailed characterizations of different film types to create camera modes that simulate the analog look of various types of both color and black and white films.ā
If reproduction is the goal, my only point would be: show me the comparisons. Of course, if Fuji (or any form or LUT or personal RAW processing) only wants to go for something that evokes the feeling of film, then everything become subjective and weāre not comparing anymore, but simply judging on personal taste. And thatās also fine. Beauty is in the eye of the beholder 
P.S. Yedlinās videoās are really great, I watched them for the first time only recently.
true, the author does seem to think so, but that quote is straight out of a Fuji brochure. I like the cameras Fuji makes, but thatās coming straight from the marketing department, not the engineers.
agreed, its a shame there arenāt better comparisons out there but for what little its worth: from all the twiddling Iāve done with my Fujifilm camera, all the results Iāve seen others get with their Fujifilm cameras, and all the actual film photographs Iāve seen in my lifetime - I think theyāve missed the mark entirely in terms of āfilm emulationā.
what they have given us a very appealing, functional and customisable system to use that gives excellent results in itās own right.
I think most people are interested in getting the feel rather than an exact reproduction. Reproduction may not be meaningful in our day, since our media and technology are different, as are our tastes.
These revivals are still significant because they bring up a nostalgia and an appreciation for history among seasoned and new photographers and admirers alike. And perhaps this interest could turn play into more serious endeavours such as archiving and reproducing the real deal.
Being that the look of Fujiās film simulations changes quite noticeably between camera generations, I donāt think actual simulation was ever their goal.
āEvoking Memoriesā is much closer to their intended purpose, I suspect. And probably for good reason, since many film stocks have severe color shifts that we would probably identify as āincorrectā today.
Another point is color gamuts, though. Some of the images in the linked article had out-of-gamut colors that clipped to a primary color. This probably didnāt happen in real film stock as āsensorā sensitivities could be matched to image dyes without white balance to mess things up. Essentially, every film defined its own color space.
That is probably the reason that I find Fujiās Velvia simulation unbearably gaudy, with greens and reds too often saturating to primary colors. Whereas real Velvia film stock could exceed the gamut of my computer monitor with ease.
1 Like