New avenues

Forums All About Photography New avenues

This topic contains 9 replies, has 2 voices, and was last updated by Profile photo of John Royle ajroyle 10 months, 1 week ago.

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • #13669
    Profile photo of John Royle
    ajroyle
    Keymaster

    Two interesting things I’ve come across recently. Firstly our favourite organisation,Google, has been looking at image sharpening and engagement and made a significant contribution it seems…

    http://www.amateurphotographer.co.uk/latest/photo-news/google-raisr-uses-machine-learning-sharpen-images-100341

    Secondly, something especially interesting to Fuji film users. People have been saying all along that the jpegs out of camera are something special. They have also been saying that Lr is not so good at interpreting the Fuji raw files. Lr has improved, but is still something short of the best converters. I recently found an interesting account by a photographer who was convinced of some special abilities of the in camera Fuji converter. In fact, despite being a skilled photoshopper he could not mimic the superior rendition of colour nor the detail quality of the Fuji jpegs. It is possible to ask the camera to convert images to jpegs after the event and I am now wondering if it is worth trying a preliminary jpegs conversion followed by detail finishing in Lr or PS.

    #13670
    Profile photo of Ian McNab
    Ian McNab
    Keymaster

    Haha! Trust Google to call the algorithm ‘Rapid and Accurate Image Super-Resolution’ so they could say it’s ‘RAISR sharp’! Not sure how generally useful it will be: it seems aimed at improving image enlargement for pictures taken with mobile devices.

    What would you have in mind as a possible workflow using the Fuji in-camera RAW converter? Certainly, there are better converters than Lightroom. Of course, I’ve been shooting mainly in BW for quite some time, and I now routinely use Fuji’s brilliant camera BW jpegs. But I have the camera set to record RAW as well as jpeg, so that I have the option of using colour where appropriate. And when I do, I use Iridient Developer to do the conversion, as it’s much better than Lightroom. In particular, the sharpening algorithms are far superior.

    Until recently, Iridient has only been available for Apple computers; but the developer is now offering a beta version of Iridient X-Transformer for Windows, that lets you convert Fuji RAFs to DNG using the Iridient algorithms.

    The Fuji in-camera RAW converter has only basic, global, adjustments. If you don’t change any of the dozen or so things you can adjust a bit, the jpeg you produce will be exactly the same as if you’d recorded RAW+jpeg in the first place. Of course, you can change things like the film simulation, white balance, exposure compensation, etc. So you could make half a dozen different jpeg variants, save them, and decide which you prefer by viewing on a computer. I’d imagine that working via the camera’s LCD display would be difficult. And the RAW converter doesn’t have facilities for local adjustments. So it’s pretty basic.

    The main use I’d see is being able to produce several copies using different film simulations.

    What do you reckon?

    #13671
    Profile photo of John Royle
    ajroyle
    Keymaster

    If the Google software is capable of achieving enlargement and corresponding detail then there are a two things that immediately come to mind; one is that it will permit more cropping and the other is that one can achieve a larger print without quality loss. I will be very interested in seeing what it can do.

    Some people are already enlarging their images using the On One software before starting work on it. The intension is not to have a bigger final print but just to enhance detail.

    Regarding the raw conversion I am just at the start of this but the details shown really interested me. Whether Iridient does as good a job I have no idea but the Fuji conversion is free of course. I am certainly going to experiment.

    #13672
    Profile photo of Ian McNab
    Ian McNab
    Keymaster

    Having read the Google Research Blog articles (linked in the AP article you pointed to), it seems possible that the AI algorithms could be incorporated into software that would run on consumer desktops. (Not sure about this: they don’t give details about how much computing power RAISR currently needs; they run it on their servers to upscale high-res images that users put into Google+ streams.)

    Do you have a link for the sources that you saw discussing the use of Fuji’s in-camera RAW converter?

    (I had a play with it this evening after reading your post. It seems pretty basic, but it is a good way of getting a variety of Fuji jpegs out of a single RAW file.)

    #13679
    Profile photo of John Royle
    ajroyle
    Keymaster

    The video about raw processing in camera is https://www.youtube.com/watch?v=rHvCkRTK4_g

    Any ideas on where to start with this would be welcome. I usually now shoot raw plus jpeg anyway, mainly to get a better image to chimp for focussing. As long as 2 years ago now I took some portraits of some friends for a sculptor to use. I happened to set the camera on raw plus jpeg and I found that the jpegs looked so superior straight out of the camera I used them.

    Incidentally, do you DSLR shooters realise the value of the WYSIWYG view you get with a mirrorless camera? It can be a bugbear when you shoot flash and you find you are plunged into darkness in the viewfinder (this is easily remedied btw) but otherwise it is a boon.

    #13680
    Profile photo of Ian McNab
    Ian McNab
    Keymaster

    When I import the RAW+JPG into LR, I only rename the files to my own scheme: I don’t let LR do any ‘import’ processing on the RAW files, because I want to do the entire RAW conversion with Iridient.

    Why does this matter? Well, it turns out that I’m able to put these untouched RAW files back onto an SD card, and reprocess them in camera to make further Fuji jpegs with different film simulations and other adjustments! And I suspect this works best if you haven’t changed the RAW (though I haven’t actually tested that yet.)

    You do have to put the RAWs back on an SD card into the right DCIM file structure that the camera is expecting, and rename them to the same filename sequence your camera is using.

    More experiments needed…

    #13681
    Profile photo of Ian McNab
    Ian McNab
    Keymaster

    OK, I’ve been having a play with this. Here’s a crop from a processed jpg, based on an OOC Fuji Mono+Y…

     

    So I put the original RAF file back into the camera  – interestingly it has to be the SAME camera, i.e. the same Fuji model – and generated a Fuji Astia jpg using the in-camera RAW converter. Here’s the result, edited in much the same way as the photo above, but using the BW image as a luminosity mask in Phototshop:

     

     

    #13682
    Profile photo of John Royle
    ajroyle
    Keymaster

    I have re-converted one picture, a portrait. I experimented with noise control and sharpening. I thought I detected a small improvement in both but I would want to do more tests to be sure of that AND be sure it was controlable. In the colouration there is clearly a chance of getting an interesting result – it is the Fuji colours which have everyone impressed. What I am looking for mainly though us suppressing the “wormy look. That textured plasticky feature which seems to be a function of the irregular pixel colour pattern that Fuji use to offset the lack of an anti-aliasing filter.

    #13683
    Profile photo of Ian McNab
    Ian McNab
    Keymaster

    What I am looking for mainly though us suppressing the “wormy look. That textured plasticky feature which seems to be a function of the irregular pixel colour pattern that Fuji use to offset the lack of an anti-aliasing filter.

    That’s a result of Lightroom’s RAW conversion algorithms. You simply avoid it by using in-camera JPGs, as the camera is using Fuji’s own conversion algorithms; that does nothing to address the LR RAW conversion problem. That’s why I use Iridient Developer: it has cracked the issues of rendering the irregular Fuji pixel layout, and converts RAF files really well.

    Iridient X-Transformer for Windows is in public beta, so you can download a free copy to try out from here:

    http://www.iridientdigital.com/products/xtransformer_download.html

    (The output files are watermarked, but you should be able to get a good idea of how good the processing is. The default sharpening algorithm is excellent, but you can alter the sharpening method if you wish.)

    #13684
    Profile photo of John Royle
    ajroyle
    Keymaster

    That is really interesting. So I now have two possible ways to get improved results; using Iridient or processing in camera. I suspect the later is going to be very awkward but free. Mind you, the number of images you need to process to the highest standard is limited, perhaps only your club entry per year, straight Lr being more than good enough for shops.

Viewing 10 posts - 1 through 10 (of 10 total)

You must be logged in to reply to this topic.