Apologies as this has probably been done to death. But I have both a Nikon APSC and Fuji camera, and I simply can’t get the Fuji scans to come out as well (I’m using the same Nikon nifty-50 lens with extension tubes on both).
I’ve read many of the posts on avoiding worms, and tried democratizing with LR enhance, DXO, and X-Transformer. I’ve tried sharpening with detail set to 0, but ultimately only disabling sharpening completely is able to avoid artifacts.
I’d like to consolidate and sell the Nikon and buy an X mount macro lens, but I don’t want to do that until I’ve got equivalent results on x-trans.
So bottom line, has anyone gotten their x-trans results to be as good as bayer, or are they just good enough?
and this points out the workaround. Use the Fuji as you deem best, export positives and finish those to taste. Use 16 bit TIFF to preserve the largest room for post-processing. Add drives if necessary for the heavy positives.
It would be very helpful if you could upload a comparison at 100% / 1:1 to show what you mean by the artefacts, or maybe provide a link to a couple of RAW files from the different cameras of the same subject/scan.
I use Fuji and rarely see anything that I could regard as ‘worms’ but I do believe that they are more likely to happen with indistinct subjects, so distant foliage or crops for example where the lens isn’t that good, or not at its optimum aperture. I think the main culprit here is likely to be the lens itself and certainly your ‘nifty fifty’ Nikon just isn’t designed for this type of work. Have you perhaps got Vlad’s Test Target?
Indeed, Vlad’s test target is in the mail. I’ll definitely get some samples once I have that. But let me see if I have any already that I can share. Random, but it seems maybe the effect is less pronounced on kodak gold 200 than on portra 400
I don’t know too much about sharpening. But I know it can be done either as part of demosaicing (dxo) or possibly after. Do you know which one lightroom does? If it’s after, I’d assume exporting (LD demosaic’d) as a tiff and then sharpening later wouldn’t help much?
There are several ways of sharpening. As you mention DxO and we’re in a LrC/NLP environment, let’s have a look at possibilities…and I’ll use the word acutance (details) because it’s all about the perception of sharpness which is increased by working on transitions.
Tools that *can* change acutance (subjective perception of sharpness)
Name scope and comment
* Sharpening ; acutance and noise
* Unsharp Mask ; acutance & balance details/noise
* Microcontrast ;
* Fine Contrast ; refined microcontrast (DxO)
* Structure ; refined microcontrast (Adobe)
* Clarity ; refined microcontrast
* De-hazing ; mixed blessing
* AI sharpening ; Adobe, Topaz et al.
All of the above work and some are easier to handle than others, but they all try to improve acutance and make the image look sharper than it really is.
DxO’s lens softness correction is slightly different. Based on measuring individual lenses, maps are created that describe how sharp the measured lens is at (any) given distance and zoom setting and how this sharpness in- or decreases over the image area, imagine an image with slightly soft edges and center and a ring of good sharpness. These maps are encoded in the DxO modules and PhotoLab can understand the stored data and increase acutance of the image with less sharpening in the ring shaped area that is sharp already.
No matter the technology, sharpening can create worms in images captured with X-Trans sensors and some products handles worm prevention better than others.
Wrapup: If, for whatever reason, you (have to) scan the films with an X-Trans camera, the way around worms is to use no sharpening before the image is available as an RGB positive copy that can be sharpened to your heart’s content. CAVEAT: Be sure to test your images for this before you waste too much money, time and drive space.
Very informative, thank you. DxO has no lens data for my contraption, and I’m pretty sure all the sliders do nothing; so that makes sense.
LR seems to give similar results for both raw and exported positives with the tools in the ‘sharpening’ section, so I suppose those come last in the pipeline. Texture/clarity/haze give very different results and much better on the exported positive, so I guess those come earlier.
Couple questions. Is it more important to have demosaicing that reduces worms (or is that not a thing?) or sharpening that is worm tolerant? If the latter, suggestions for best tools?
Final question: if I use a tool like upscayl, is it better to do no sharpening at all before upscaling and then sharpen? or sharpen first?
I think that Nate uses (used?) an an X-T2, so you’d think he might have identified this as a problem. It will be interesting to see your examples but I think that the more detail you give the RAW processing software the less likely it is to make things up (worms). Very useful test though.
Fine for web IMO but if I enlarge and sharpen for print then it looks not so good (although I haven’t gone so far as to experiment with ordering prints yet).
Obviously you’ve ordered the VTT in order to establish how good your lens is, find the optimum aperture, check that alignment is perfect etc. and it’s great for that. I’m wondering now if there is a part of it that might show up Fuji worms, or at least different ways of RAW processing, and different sensors. Given that everyone can buy one and share results it would be good if it could and we could all compare results, but of course it’s basically a high resolution target on extraordinarily fine grain film so maybe not, in its present form anyway. Perhaps Vlad could include a fudgy test area.
Ended up selling my Fuji system for a Nikon one as I was constantly unhappy with the weird X Trans rendering. It’s not night and day but am much happier even with the original Z6 and Sigma 105 (which was also the lens I used on the XT3). I mainly use the camera for scanning film so switching made sense. Miss the Fuji a bit but I honestly don’t really shoot digital photos now … or if I do they are on my phone.
Would have been interesting. You did say that you were using a Nikon ‘nifty fifty’ with an extension tube rather than a Sigma 105 though. Sigma should have been OK if a bit long for APS-C.
Ah yes. Ended up doing a range of things. You are right, I originally had the 50mm Nikkor. Must have got the 105 with the Z6. My memory is getting worse
Vlad is in! Here’s a corner from the nikon 50 at f5.6. sharpening 40, radius 1, 0 detail, no noise reduction, demosaic’d in lightroom not using enhance. IDK how good most folks are getting their test targets, but to me this is at least not bad
Thanks for posting, but it is the difference between the centre and the corners that matter and separate good lenses from bad ones. At 24MP with Vlad’s excellent target you can expect to define Group 0 Element 2, but you should also be able to define it in the corners with a really good lens. With my Fuji X-T2 that’s what I get with the 75MM Apo-Rodagon f4.5 2x, but the (much easier to find) 80mm f4 Schneider Componon-S is very close.
Vlad did his own comparison on 30MP full frame between the same 75mm f4.5 Apo-Rodagon 2x and the very well regarded 70mm Sigma ART Macro:
Mathematically doubling the Megapixels should allow you to define 3 elements higher.
The results with the 50 aren’t good (obviously, we don’t expect much from this lens on extension tubes). But I thought the question was if the overall setup was giving enough detail to the software to avoid worms. LMK if I have the wrong take-away, but at least center frame it seems like we should be OK there. Certainly the test target shows no worms on the black and white areas. I can however see worms on the gamma (or whatever it’s actually called) area with sharpening turned up (still 0 detail).
So what does that tell us? We can expect worms on large flat textures like skies unless we’re able to resolve individual grains to give the software something to work with?
Well you didn’t mention worms when you posted your first results from Vlad’s Test Target, and you didn’t include the greyscale area which you now see worms in. Probably the most useful comparison to make now that you have the VTT would be with your Nikon.
