The Fujifilm X-H2 High ISO Noise vs the X-T2

Yidahoo wrote:

Thanks everyone for making this a very interesting discussion. I would like to point out that nowhere in the video have I claimed to be inexperienced with RAW files. I have been shooting and processing RAW for nearly 20 years.

The idea for this video was simply to compare like for like with absolutely no processing. I plan to do another video where I take the same files and process them into usable images in Lightroom, using exactly the same editing settings and applying the same amount of LR noise reduction in order to see how they stack up.

I don't own any third party noise reduction software as I don't shoot high ISO very often.

There is another thread where certain folks are claiming Noise doesn't matter. Well it certainly does, regardless of which Brand or Sensor one is using. Some are claiming it doesn't matter because of modern noise reduction software. Some are claiming many things about it.

With so many folks with claims how Noise doesn't matter, it's real funny how so many do indeed own, third party noise reduction software.

Most times that I do shoot at high ISO is mostly for testing reasons. To see just how far I can push the camera for my tolerance levels. I normal try to stay as close to base ISO, as humanly possible given what gear I am using an or the subject matter. The current conversations about Noise is mostly centered around the higher MP cameras, be that cropped or FF.

“The most critical roll in determining noise is the photon flux. In reality all comparisons should be scaled to a standard photon flux to make a valid comparison”

Broadcast video cameras are spcified this way. Snr is measured with a stadard target illuminated at 2000 lux

baobob wrote:

There is a difference between it does not matter and it is not an issue. IMO it does matter but it is not an issue with the XH2 sensor provided in dim light you ETTR, AF is perfect, and use a lens with a good contrast to have a high SNR. The 40mpx sensor is not forgiving approximate settings. As usual with new specs there is a learning curve to obtain optimal results.

When needed very high ISO will allow to catch the scene at the cost of some loss in IQ Denoising programs are just to help to maintain usability of the shot. It is clear that it is better to set camera in order to minimize the level of noise.

When I got my XH2 alongside the XH2s I fouynd it very noisy. Fortunately there are here participants who had already experience with the camera and there comments and advices helped me to obtain optimal results with the XH2.

At the end it is now for me no longer an issue.

Clearly the differences are personnel to each individual. That I am perfectly okay with. The issue I have is when anyone wants to tell anyone else what levels of comfort they should have with their own images.

I don't care how sharp, what noise levels anyone wants with their own results. But I have a huge problem anyone telling me what levels don't matter or isn't a issue for me.

By the way, I totally didn't agree with the yTuber Omar about the difference in IQ between the 2 sensors, 40MP and 26MP. I don't like yTubers doing those types of tests, then spreading it around as the Golden Goose that laid a Golden Egg.

I also believe that for the most part, Higher MP camera give more detail at the cost of more noise. All things else being equal. So I lean towards higher count MP cameras moving forward.

Thank you for a very interesting and useful video link. The follow-up on processed photos will be even more interesting and I eagerly await it as it is the processed images which are critical in practice.

However, the key to the whole problem is to find a solution which gives more than acceptable results with the minimum of expenditure and if you can find a way to make any image look up to your standards at the size you print or view without upgrading to FF or MF then this should be thoroughly explored first in my view.

Jerry has mentioned Topaz Denoise AI and this is a really excellent approach which gives better results than LR.

There are other avenues well worth exploring and I’ve found that Deep Prime, as used in DxO’s Photolab 6 (as a LR plugin), produces outstanding results as it maintains sharpness and can virtually banish noise, especially on very high ISO images such as ISO 3200 and 6400. (The cheaper version, DxO pure RAW) doesn’t have the important flexibility of changing the amount noise reduction applied at different ISOs and, while worthwhile, it isn’t as good as the more flexible version in PL6 which I reckon can give up to an extra 2 stops!

DxO often has free trials and it is really well worth trying it out. It’s not dirt cheap but it is hugely less expensive than upgrading a camera or a whole system (as is Topaz).

Perhaps as important though is the need to maximise S/N by exposing to the right!

However, I regret to say that giving in to GAS is often far more exciting in the short term than upgrading the S/W!

Disclaimer: I have no relationship of any sort with DxO but you’ll have gathered I find it invaluable.

Phil

Because it is a dam sight cheaper than upgrading to FF and is often more than good enough – what more would you want?

Phil

Phil1 wrote:

Because it is a dam sight cheaper than upgrading to FF and is often more than good enough – what more would you want?

Phil

I want better than "more than good enough".

As to it being that much cheaper as you suggest, I don't think so. It totally depends on your shooting needs, but some FF options provide many affordable, quality third-party lenses. Plus some very good FF cameras currently cost the same or less than the X-T5/XH2.

Kaoticphoto wrote:

Maybe I'm not proficient with the denoising tools yet, but I think people are putting too much hope in their ability to make high ISO files from this new sensor much better. From what I've seen so far, anything above ISO 4000-5000 is worse than what the X-T2 produced, let alone the A7III. And these are 5-6 years old sensors. I can't help but think these new sensors are rather disappointing for low-light photography.

`these are 6400 iso, processed in same way, resized at same size, X-H2 and A7III

https://www.kaotic.it/trash/A7III_6400.jpg

https://www.kaotic.it/trash/XH2_6400.jpg

Thanks for these, I had missed your reply. Whatever you did to the A7III file, it made it slightly worse in several of the areas where it looked to retain more detail. However, note that it's the unprocessed RAW (and resulting JPG) looks noticeably better than that on my settings.

The XH2 looks pretty much like what I get from processing in Lr and Topaz Denoise.

All in all, fairly close, but that's my point. The A7III, despite being a much older, and, right now, cheaper camera (at least where I live) produces at least equal (in my view slightly better) results without any additional processing.

Anyway, each of us knows what is good enough for them, and whether the trade off is worth it.

Tbh, Fuji IQ hasn't changed much since the first X-Trans sensor. My XE-1 gives the same IQ, even the noise, as my XT-2 and XH-2 at least up to ISO 6400. Maybe only the dynamic range is better in the newer sensors.