X-Trans Experiment

[Adzman808]

Thanks, Adam.

I have the raw conversion nailed (ish. I hate Silkypix and have no desire to go through the agony of learning how to use it properly for my secondary camera, but I can just about use it), but if a particularly watercoloured shot comes up, I will hand it over for you to have a go at. I do use noise reduction, using Noise Ninja, and sharpening, using a frequency separation method. So, I do both with care. However, there are just areas of tone and colour where there is nothing, so denoising and sharpening can't really make it worse. I will see if I can find a detail of a shot that shows you what I mean.

No problem, but let's start our own thread for that one, lest we hijack this one all the way to the moon!

[Max von MeiselMaus]

Adam cross posted, so already in this thread. As we are talking about the X-Trans sensor, it is all good.

OK. Here we go. I tried to find two pictures that were as close as I could get in terms of settings, which isn't that close. Both are 100% crops of stonework on an overcast day. The settings are all over the place as I had only just started to take outdoor photos and was still finding my way (still am, but am a little more sensible in my choice of settings).

These are pretty much straight out the camera, with a little bit of levels on each, to brighten them.

The top is with a Canon 60D, f/2.8, ISO400, 1/200.
The bottom is with the Fujifilm X-E1, f/4, ISO800, 1/500



These are the same, with a little noise reduction and sharpening. If I had done this properly, I would have processed them separately rather than together, but I couldn't be bothered.



So, the difference is subtle and is actually a lot less than I was expecting at this size. However, the Fujifilm offering has a slight over-processed look to it. It is less gritty and looks smoothed out. This is the watercolour effect. When you see it all over the photo, it is still noticeable.

What would be a real test is for people to post images from the old X-Trans sensor and some from another, unprocessed, and to see if we can spot the X-Trans photos. It would probably be more difficult than we would think.

[Black Pearl]

Absolute pixel level detail is only one tiny part of photography - relevant if you pixel peep or print gigantic images then study them up close or like to fiddle with minutiae.

For me (I share images online and occasionally print at moderate sizes) dynamic range, colour rendition, iso control, the way the camera feels in the hand, the lenses available, the fps, the buffer capacity, the size, the weight, the hard to put into words but make a difference in an irrational way things are all far more important reasons why I buy and use a camera. I found all of those things in the X-T1 and the more I use it the more I am falling in love with it. I miss the 51 point AF system from my D300s and I miss the weather resistant body/lenses from the K30 (I drenched them regularly) but not enough to regret dumping them all for the kit I have now. I am falling back in love with the process of simply going out and using a camera that feels amazing, has tactile dials, is predictable and makes me happy to own it. The results for me are better than I had and while that matters its not the be all and end all - I just love using it.

[Adzman808]

The fact the Fuji shot is iso800 makes me wonder of you were using the DR Auto (or DR 400) mode?

If so, (IMO) you need to be turning that off. It does your raw files no good at all!

DR expansion on Fuji (& probably other cameras, but I don't know) works by underexposing the shot by 1 (DR200) or 2 (DR400) full stops.

This is not good for a sensor who's widest dynamic range is at base ISO

When we make a SOOC jpeg using DR, the underexposure protects the highlights, then the internal convertor SELECTIVELY boosts the tone curve to bring up the shadows

When we take a raw file from the camera after using DR expansion the step that boosts the shadows is skipped, so you end up with a raw that is 1 or 2 stops underexposed, this leads to all the 'problems' that you'd expect from under exposing, namely, noise, lack of detail, smeared detail.

Some raw convertors are better equipped to deal with this (apparently, according to the chap that writes all the Fuji books - Rico Pfirstinger) but iirc SilkyPix is not one of them.

I appreciate you can lift the shadows in post, but when you've unnecessarily thrown away some of the native DR of the sensor, you're just amplifying noise (SNR)

My sincerest and unreserved apologies if your shot was taken at ISO800, because it warranted it, and no DR expansion was used...

However, what I've written is true, and IMO worth posting in a thread about X-Trans, even it doesn't apply to your posted examples

By contrast, if one leaves DR off (on Fuji this means set DR100) and still the camera deems ISO800 to be correct, then at least the signal amplification is applied to the whole image, and not just to dark and mid tones selectively.

There's pages and pages written about this on the 'net, but I'd like to think that I sum it up rather concisely...? YMMV!!

On X-Trans, digital amplification of signal occurs at approx. ISO>1600. Try to avoid this IMO..

There's some school of thought, which I don't pretend to fully understand, amongst die hard Fuji jpeg shooters, where you set ISO400 and DR200 in a scenario where the camera wants to use ISO200, then use the EV wheel to dial back a stop to prevent the underexposure (turn the EV dial to a negative value and shutter speed goes up) so that the exposure triangle is still consistent with an ISO200 exposure, then the in camera algorithm will protect the highlights and give you a nice "high key" SOOC jpeg

But for my HUMBLEST opinion, if you're prepared to bugger about that much to make SOOC jpegs, then jez.. just shoot raw and bugger about at home, on the PC/Mac, once the kid(s) are in bed, the OH is watching TV and you've perhaps got a glass of Cognac on the go while you review the day's shots

[Max von MeiselMaus]

That's interesting, Adam. I tend not to use anything I don't understand and certainly not any suspicious auto settings. Sounds as though I am not missing much with this one.

Both shots were in aperture priority and I pushed the ISO up on that one as it was dark and I wanted a fast shutter speed. As it happens, and with experience, I realise that 1/500 is WAAAAAY too fast for what I needed (those buildings aren't jumping around like maniacs). But these were amongst my first outdoor shots, so lots to learn. I now tend to set the X-E1 to auto ISO, 1/125 (I have shakey hands and could probably go a lot lower than that. I need to experiment) and as small an aperture as I can afford to get a decent exposure without going into ISO Ridiculous.

Robin, yes. These are very nice cameras and I now take mine out with me on the off chance there might be something worth photographing. Any outing with the 60D was an exercise in back ache. I do, however, find that the overall photos still have that watercolour effect, in fact more than the 100% crops would suggest. When I am shooting something textured or want a gritty feel, it can be frustrating. However, the advantages of the camera outweigh that.

[Pictus]

X-Trans RAW conversion tests
http://www.fujivsfuji.com/best-xtrans-raw-converter/
http://blog.thomasfitzgeraldphotogra...ans-processing

I like Photo Ninja, but to use with a X-Rite PassPort to create custom light profile, it is a kind
of "super white balance" that takes into consideration not only the neutrals but also all colored patches.
One example http://www.ultravioletphotography.co...light-profile/
RawTherapee is FREE and can get good details too, but a bit complicated...

[Black Pearl]

There is talk and fairly compelling info on the net that using 800iso and DR400 increases the DR - EVEN when shooting raw. The reason being Fuji apply (bake in) quite a bit of processing to the raw data and part of the DR trickery makes its way to the raw files. I haven't tried this and to be fair I'd far rather stick with using base iso but what I have found is the way you can maul the files in LR and still get a very good output can lead to some astonishing end images.