Holy cow.
I thought the people complaining about bad demosaicing on Fuji RAW files were just being pixelprudes, but this is ridiculous.
This picture shows darktable failing miserably at proper rendering of a basic Fujifilm RAF. Even this basic, flat gray wall now looks like glittery rainbow clown snot.
How is this the meme camera system that everyone talks about if all you can do with it is shoot JPEG?
How do you properly post-process RAFs?
please someone tell me that this isn't the nasty punchline to some sick joke
>>2860746
>Fujifilm RAWs
>Darktable
Found your problem
>>2860748
whats the proper way to do it?
>>2860749
Literally anything else.
Raw processing is very application specific, developers of raw software are constantly trying to differentiate their products by producing better renditions of different raw files, all you've found out is that Darktable is terrible for fuji. I can tell you that lightroom is quite good for x-trans nowadays. You can find a great comparison here and then choose the best compromise of ease of use/versitility/image quality there after: http://www.fujivsfuji.com/best-xtrans-raw-converter/
You can use Lightroom 5.4 in Linux under wine.
>>2860824
Lightroom is also really bad at handling RAFs.
>>2860834
It's possible, but unless you know of a good RAF to DNG converter, I think it will suffer from the same problem.
The problem with Fujifilm RAFs is that they reflect the irregular subpixel color pattern on Fuji sensors, so they need a different demosaicing algorithm. Bad demosaicing is where the clownbarf colors are coming from.
When the demosaicing is done properly though, there is virtually no chroma noise. This is one of the things that makes Fuji's out-of-camera JPEGs so appealing.
----
The guys on the internet say Irident Developer is one of the only softwares who's developer has figured out the proper demosaicing technique. But I don't want to shill out 99 of my hard earned US dollars to some shady no-name software harem that will probably go defunct in the next 5 years.
>>2860746
You need to do the specific demosaic configuration for Fuji RAW, it is on the Darktable site, then save it as a preset and set it as default preset.
Darktable can handle RAFs well, not as good as Lightroom or Photoninja, but better then in your pic.
This is nothing more than user error.
>>2860746
And you wonder why every fuji user harks on ovwr "muh jpg presets"
Or how none of them know how to use editing software correctly
Or why every photo from a fuji has that same confused muddy colouring
They're toy cameras for those not really interested in the technical side of developing a photo and just want to click a button and look "cool"
>>2860981
>>2860746
>How do you properly post-process RAFs?
by using a different raw converter
if you want to stick to linux, use rawtherapee
>>2860969
please tell me this isn't the post you're referring to:
https://www.darktable.org/2014/08/using-x-trans-cameras-with-darktable/
because I have read it and if you see in the screenshot, I have already applied what the post says is the overkill-quality demosaicing algorithm
>>2860981
im sick of your stupid 'toy camera' remark that you regurgitate inevery fuji thread. I think I've seen it like, thrice now. It's not funny or enlightening. it's just irritating and i wish you would just get banned. Totally uninformative and unhelpful shitpost.
At least it bumps the thread.
>>2860981
>A few loud idiots mocked by a few other loud idiots
Someone makes a thread asking for help and instead of helpful comments you spew this bullshit memetic hate speech.
This is why /p/ went to shit.
Also >>2860746
https://www.darktable.org/2014/08/using-x-trans-cameras-with-darktable/
If it doesn't work then try Lightroom. If you are bound to Linux too much then there is the alternative AfterShot Pro 2 native Linux version.
>>2860992
Oh. Then it's time to try other software.
See if AfterShot Pro 2 works, it should have some photoninja code integrated.
>>2860997
>>2860993
thanks. i guess the next step is to try some windows raw software in my vm.
attached is the fujifilm jpg vs rawtherapee (3-pass, 5x smoothing) comparison, which does marginally better, but not enough to stop looking for alternatives. The chroma noise now looks like an overlayed plasma cloud, and edges still look like they have chromatic aberrations.
>>2861012
I don't like Rawtherapee for my bayer RAWs either.
ASP2 is linux native, the demo version is just a time limited full version so give that a try next.
Once you process your RAFs youre only looking at m43 quality photos. Kinda sad when you really want that hipster aesthetic but only fujifilm makes cameras that clunky looking.
>>2860992
Thats the first time I've called fuji toy cameras on here, its been used a lot, mainly because they're joke toy cameras.
Also, reel in the butthurt isi.
Daily reminder that the summer cancer finally breached /p/.
The samefag above is the proof.
>>2861012
>literally a picture of a dumpster
I forgot for a second I was on /p/.
>>2861026
He was taking a landscape but due to Fuji's in body image processing they all came out like this.
>>2861026
It's where Fuji belongs.
>>2861026
>>2861029
>>2861030
guilty as charged
[EXIF data available. Click here to show/hide.]
Camera-Specific Properties: Equipment Make FUJIFILM Camera Model X-T10 Camera Software GIMP 2.8.14 Maximum Lens Aperture f/1.4 Sensing Method One-Chip Color Area Focal Length (35mm Equiv) 35 mm Maker Note Version 0130 Image-Specific Properties: Image Orientation Top, Left-Hand Image Created 2016:06:10 15:58:56 Exposure Time 1/500 sec F-Number f/8.0 Exposure Program Manual ISO Speed Rating 400 Lens Aperture f/8.0 Brightness 9.0 EV Exposure Bias 0 EV Metering Mode Pattern Light Source Unknown Flash No Flash, Compulsory Focal Length 23.00 mm Color Space Information sRGB Image Width 1000 Image Height 520 Rendering Normal Exposure Mode Manual White Balance Auto Scene Capture Type Standard Sharpness Normal Subject Distance Range Unknown Sharpness Normal White Balance Auto Chroma Saturation Normal Flash Mode Off Focus Mode Auto Slow Synchro Mode Off Picture Mode Manual Exposure Continuous/Bracketing Mode On Blur Status OK Focus Status OK Auto Exposure Status OK
>>2861032
It's a straight-on shot of a garbage bin.
I'm still not entirely sure of /p/ likes this shit for reals or ironically.
I don't myself like it at all.
>>2861012
>The chroma noise now looks like an overlayed plasma cloud
And you do not want to reduce chroma noise, right?
And you do not want to use false colour suppression? (cannot understand whether "3 pass" refers to to false colour suppression)
>>2861041
IDK what '3-pass' means for rawtherapee.
The "false color supression steps" slider is set to its max, 5. It looks even more abysmal if I move it down.
>>2861041
>>2861046
>cannot understand whether "3 pass" refers to to false colour suppression
>Sets the number of median filter passes applied to suppress demosaicing artifacts when applying the demosaicing algorithm. False colors (speckles) could be introduced during the demosaicing phase where very fine detail is resolved. False color suppression is similar to color smoothing. The luminance channel is not affected by this suppression.
False colors are generally more apparent in images from cameras without anti-aliasing filters. Note that it is foremost the chosen demosaicing algorithm which is the deciding factor in how prominent will be the false color problem with which you will have to deal. In some situations it may be better to change the demosaicing algorithm than to enable false color suppression, as the latter reduces color resolution.
>>2860944
Ah righto :( that's a pretty idiotic move by Fuji, to have an encoding for raw format that's so inaccessible. I understand that noise-free stuff but they could've at least thrown in an inbuilt DNG shooting option for those who don't care about ISO >6400. I don't even shoot Fuji so it's none of my concern, neither do I care, but as an opinion - it's rather inconvenient studio work and landscape to not work with the raw file.
>>2861050
I have just opened RawTherapee 4.2.699 and it looks nowhere near as bad as OP suggests.
4.2.0 is few years old even though is the latest stable version.
I use Capture One, no problems at all.
Don't use Lightroom.
>>2860944
>The guys on the internet say Irident Developer is one of the only softwares who's developer has figured out the proper demosaicing technique.
Why the fuck does Fuji not publish the proper algorithms?
Demosaicing should be quite easy.
>>2860746
> installed Darktable on my distro
> tried it out for a few days
> doesn't understand color profiles
> meh
> try to merge HDR
> image is purple
> sudo apt-get remove darktable
> boot into windows and install a cracked version of lightroom.
I don't honestly know why anyone would use Darktable.
>>2861156
>install trial version of photoshop
>trying to merge pano
>check "correct spherical aberration"
>the resulting canvas has a form of exaggerated smile
>cannot make this shit up
>trying top open Sigma X3F files
>recent ACR
>not supported
>nigga wat
>not a single untwisted/linear profile inside distribution
>the resulting curves are piece'o'shit (should be perfectly straight because I disabled all adjustments)
I don't honestly know why anyone would use Adobe software for image processing.
>>2861156
> HDR
that's your cue to leave
>>2861310
You sound like a faggot.
>>2861099
They have. Years ago. Don't know why Adobe still fails so much. Might be the RAW engine itself, being so optimized for the ordinary Bayer color grid, so Fujis Xtrans screws it over.
>>2860746
When I had my Fuji X-E1 I could always count on Iridient Developer, but I've heard good of Capture One and Photo Ninja as well.
Neither being available for Linux as far as I know though.
>>2861729
Is irident worth the price? I'm a poor college student and dont want to skip too more ramen dinners than i have to
>>2860746
I use the silkypix backend converter on the fuji website to batch convert to tif, then edit in PS. I hear capture one does alright too, though im just trying to convert the files
>>2861728
It definitely seems like this is the result of some Adobe technical debt, which is a real bummer, as I generally like the ecosystem they offer.
>>2861731
Does this actually work? Because, shit that's great if so. Don't they do the film simulations in there too? Does that bake into the TIF? This is another place where Adobe just can't get it right. It's not even close most times. Not that I _need_ film simulation, it is a nice place to start though.
>>2861731
I use the silkypix raw converter too. It's a bit wierd and limited but it works.
>>2861058
How stable is the 2.6?
For RAF files, LR is ass, even the latest version.
Irident is the best I've used, but is OS X only. Capture 1 Pro seems like a good alternative, but I'm used to LR's interface from when I had a DSLR, and C1 does have a steep learning curve.
I've read good things about PhotoNinja and SilkyPix.
>>2861058
I'm trying out RT 4.2.922 right now and it is literally just as bad. Here's the RAF
GE TT com slash 3gnJrNb2
>>2862282
rather, ge dot tt dot com slash (...)
I import pics into LR for cataloging. Then individually load up the picks in silkypix and import those tiffs into LR and do the usual LR/PS editing and export to jpeg.
>>2861055
Wouldn't matter if it were DNG. The problem is the X-Trans pattern, not the file format.
just use the fucking pirated capture one pro or photoninja. jesus christ its like torrents dont even exist.
>>2862297
>>2862283
>>2862282
>>2862360
Well that's a shame. A lot worse in that case.
>>2862360
>Wouldn't matter if it were DNG.
Strictly speaking X-Trans RAWs may be packed into DNG too if only Adobe adds corresponding tag values.
So no, no problem in being RAF too.
>>2862359
this
>>2861730
I'd say so. But it's only a raw converter. No library etc.
>>2861730
Do you need superb fine detail for large prints? Then yes. Do you print small, or present on the internet? No.
>>2862396
>>2862297
>>2862282
I just tried 4.2.1005, results: https://filebin.net/scyrc8liaunr51w1
Please post your output from SilkyPix. I wonder about whether it may get any better.
__________
By the way, what Fuji did is they sacrificed chroma resolution even more than it is already sacrificed in popular RGGB sensors. In RGGB sensors there is a at most 1 pixel between each pair of adjacent red and blue pixels. In X-Trans the spacing between adjacent red and blue pixels is from 1 to 2 and that is quite much because of how requring the modern users of cameras are.
>>2862519
Starting from bad stock makes a lossy jpeg look even worse.
>>2861086
2nd
>>2860749
fujifilm software.
>>2862935
ding ding ding we have a winner.
Left is out of camera, right is the free silkypix software on Fujifilm's website.
"RAW FILE CONVERTER EX 2.0 powered by SILKYPIX" out of the box handles demosaicing without a problem and seems to retain detail well.
As a plus it will also apply Fuji's in camera film presets to the rafs, but IDK if it is precisely the same as what the camera does. I might test this later.
As a con, it is extensionless, feature-poor, unmaintained, and the interface feels like it came from a game demo CD from a 2003 cereal box
>>2863623
Yeah, obviously better. I could not think that X-Trans is such a huge problem
Fujifilm are fags because they do not release the demosaicing algorythm.
>>2863723
>>2863623
I doubt it's the same as what the camera does. I'd guess Fuji does their processing on each channel before demosaic.
But why is Fujifilm responsible for everyone being retards? They're doing their processing on a shitty camera chip, when every raw processing program has a full desktop CPU. Adobe should be destroying Fuji cameras at rendering.
>>2865817
device designers and desktop programmers have very different philosophies when it comes to designing code.
for 99% of applications, the advent of modern, fast computers has caused desktop programmers to become lazy in general. instead of
> "i have more processing power so now i can make the computer answer harder problems in the same amount of time"
its now:
> "the computer is fast enough that instead of working really hard-to-program fast code, i can save time and effort by just writing lazy, more abstract code."
device manufacturers to some extent still use the first philosophy because they have to cram the best features into the cheapest (worst computer) package possible that can get the job done.
plus it's common knowlege that the source code behind anything from Adobe is a giant pukestain
Also, im looking into this seriously now. Virtually all the free RAW editors for XTrans use dcraw to do the actual decoding, which uses a demosaicing algo designed by Frank M (complicated last name i can't remember). this is what causes the nasty artifacts.
you can see the code theyre using for demosaicing by opening up the latest version of dcraw.c and ctrl+f RAF a few times. it's pretty well commented
If someone figures out a better algorithm and gets their patch applied to dcraw, we will see all the open source lighttable softwares eventually adopt it and get better.
i might start putting some thought into this...
>>2860746
>GNOME 3
>>2865872
>caused desktop programmers to become lazy in general
You say it as if it's a bad thing.
Development time is the most expensive time of all.
And there is a huge global shortage of software developers.
>>2863623
Silkypix is pretty much the best quality converter you can find for Fuji raws, shame that doesn't have nearly as much processing options as Adobe tools (which still can't match OOC JPG after years of improvements). But hey, at least you're not forced to use Sigma Photo Pro.
>>2865872
"More abstract" code is a good thing: it increases code reusability, facilitates error identification and automated testing, speeds up implementation and documentation time, prevents human errors and are HARDER to design than less abstract code, for the exact same result.
Sure, it may be slower, but the benefits far outweigh the performance hit.
It also has nothing to do with what you're talking about. Adobe's processing of X-Trans sensors suck because they've invested years researching a demosaicing algorithm for conventional Bayer sensors, and there's just not enough demand for Fuji processing to justify the same amount of effort (or to justify paying Fuji for their algorithm).
>>2861730
Iridient Developer is absolutely worth the $100 you pay for it. This is how I use it on my workflow:
Choose and import RAF files into LR
Use Open Directly plug-in to open the RAF files in Iridient
Make highlight, shadow, exposure, curve, sharpening, and noise adjustments in Iridient
Export as TIFFs
Import TIFFs into LR, do color adjustment and cropping there, export as jpg
1000% better results than using Lightroom alone
>>2865914
>Development time is the most expensive time of all.
Fuck no. The most expensive time is the time which users spend on trying to fix rapidly developed software - doing the work which they were not teached to do.
>And there is a huge global shortage of software developers.
Cannot argue becaue have no clue.
>>2865817
>But why is Fujifilm responsible for everyone being retards?
They are not responsible for anybody at all. It's just that the lack of software limits the use of their cameras. They are hurting both users and selves.
