[luftfluss]

Through a different lens: why we're changing the way we process lens images: Digital Photography Review

Simple explanation
In simple terms, we'll now be applying all manufacturer lens corrections when we process images in lens sample galleries, since we believe that's how most people will experience the results of modern lenses, most often.

[gatorguy]

Through a different lens: why we're changing the way we process lens images: Digital Photography Review

That seems like a "Duh!" statement. I can think of no rational reason not to base a review of a camera manufacturer's first-party lens specifically and most 3rd-party lenses too with the lens corrections enabled before this. Who turns those off? The OOC image is what matters anyway, AFAIAC it should be raw for testing purposes but that's a different aspect of it.

Kudos to them for finally acknowledging change was needed.

[biz-engineer]

Kudos to them for finally acknowledging change was needed.

I'm not sure it is a good thing to provide us with lens characteristics after correction or else MTF charts should also be provided after correction.

For example, one reason I don't apply distortion correction is because it lower the resolution of a lens.

Let's take two wide angle lenses:
- Lens A has the best sharpness of all, but has a lot of distortion (allowing more distortion makes wide angle lens design easier)
- Lens B is average sharp but has no distortion (the optical design is closer to a true rectilinear design)

Turns out Lens A gets a better score than lens B , but after distortion correction is applied to A it actually isn't as good as Lens B.

For me it's better to compare the lenses without any software correction, because the less amount of software correction has to be applied the better. Each test should tell if correction is applied or not.

[house]

Seems easy enough to show and discuss both as lenstip does it.

The proposed method is better than current one though. Only problem is that lenses without adobe profiles will suffer unless its all sooc that is.

But if its sooc its meaningless for raw shooters because software handled things differently. Hmm they should discuss both behaviours.

[Michail_P]

Testing the lenses after correction is ok, because most users will actually correct at least distortion and CA in pp. There are also many users shooting jpeg, but when it comes to detail it really should be raw and untouched. That’s a better indicator of the optics’ quality.

[Serkevan]

But if its sooc its meaningless for raw shooters because software handled things differently. Hmm they should discuss both behaviours.

And have them work??
Oh, you . Jokes aside, even though they had corrections off, they kept using the "default" JPG profiles, and judged IQ on those. That isn't a very good methodology when each brand has a distinct "default JPG" engine.

[stevebrot]

Given that with some brands, the corrections amount to AI re-imagining of sensor output, this statement amounts to a bow to certain prominent players where such is applied to all JPEG output. FWIW, the majority of Pentax lenses are not supported by manufacturer's correction profiles.


Steve

[luftfluss]

Given that with some brands, the corrections amount to AI re-imagining of sensor output, this statement amounts to a bow to certain prominent players where such is applied to all JPEG output. FWIW, the majority of Pentax lenses are not supported by manufacturer's correction profiles.


Steve

Not just JPEG - some lens corrections are embedded in RAW, as well.

[Paul the Sunman]

Not just JPEG - some lens corrections are embedded in RAW, as well.

... but can be overridden.

I have all lens correction turned off on my K1 and distortion and vignetting correction turned off on my K3iii. I only shoot Raw, and enable them in post processing occasionally, but I would rather have that choice as an opt-in rather than an opt-out.

[cowlitzJim]

One reason I've sought out TTArtisan lenses in the past is that they give excellent results on μ43/fuji with no corrections applied. That's also why I returned to Pentax - to use lenses in a system where such corrections are optional.

[biz-engineer]

Good and bad. Good because it's good to have image output as good as possible with minimal work. Bad because evaluating lenses after correction give advantage to brands doing it , and encourage every brand to follow that path and then we'll not know if the glass we buy is really good or a software corrected dud. Also a disadvantage for third party lenses uncorrected by camera. e.g compare a Tamron 24-70 2.8 (uncorrected) to a Nikon 24-70 (software corrected), the Nikon look a lot better but in reality it's probably only marginally better than a software corrected Tamron. Lens tests should provide corrected and uncorrected results IMO.

[stevebrot]

Not just JPEG - some lens corrections are embedded in RAW, as well.

Please provide documentation of this in Pentaxland. I have done a fair amount of forensics on the subject and own a few of the lenses with known correction profiles and have not found evidence of such.


Steve

[StiffLegged]

… evaluating lenses after correction give advantage to brands doing it , and encourage every brand to follow that path and then we'll not know if the glass we buy is really good or a software corrected dud.

The software might be able to compensate for vignetting in corners, distortion and aberrations, but it can’t bring back details the lens couldn’t resolve.

[Michaelina2]

Given the self-correcting nature of AI, and stepping into the (say) near future…

If in-camera image correction is now part of a lens score and I build camera+lens kits, I would want to know why EVERY lens I build from now on, failed to get DRP's “Gold” award with my camera body.

Eh…?

Cheers… M

[Rondec]

I'm not a big fan of this (not that I'm a big fan of anything DP Review does). The issue, as I see it, is that we need to know how much correction has to be done to an image. If there is 5 percent distortion and 3 EVs of vignetting in the corners of images from a wide angle lens, even if the corrected image looks pretty good, it has to affect your ability to post process that image. You will have less resolution and more noise in the corners.

Lens corrections can be helpful, but to my way of thinking, it is better to start with a well corrected optic than to think you can take something that is sub-optimal and massage it to look decent in post.