Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
First test of the Tamron 20 ƒ/2.8 Di III OSD M 1:2
#11
(01-02-2020, 07:16 PM)faint Wrote:
(01-01-2020, 12:01 PM)Rover Wrote: https://www.ephotozine.com/article/tamro...erformance

8.5% native barrel distortion is... wild. Smile I wonder what kind of value is considered the borderline of the fisheye territory?

There is some tradeoff between distortion and MFT resolution values for wide angle lenses.

I wonder if it is OK to leave distortion less corrected when every other parameter is more or less optimized, and leave the end user to decide what distortion correction to apply during post-processing. Obviously, landscape people will want better edge sharness, architectural people will look for straight lines.

Well, I think that's exactly what has been done lately by most producers? Compare the two versions of the Sigma 12-24/4.5-5.6: the first one was not as sharp as the second but had a lot less linear distortion; for the second version the inverse was true. Many comparatively recent lenses, even in the DSLR lands, have been made less corrected for distortion but with better overall sharpness than the older versions. The likes of the Nikkor 24-85 VR comes to mind immediately. And in most of the mirrorless realm, it's really gone laissez-faire (Olympus, Fuji, or that Sony 18-105 with the pincushion measuring upwards of 6% at some FLs...)
#12
(01-03-2020, 01:48 PM)Rover Wrote: Well, I think that's exactly what has been done lately by most producers? Compare the two versions of the Sigma 12-24/4.5-5.6: the first one was not as sharp as the second but had a lot less linear distortion; for the second version the inverse was true. Many comparatively recent lenses, even in the DSLR lands, have been made less corrected for distortion but with better overall sharpness than the older versions. The likes of the Nikkor 24-85 VR comes to mind immediately. And in most of the mirrorless realm, it's really gone laissez-faire (Olympus, Fuji, or that Sony 18-105 with the pincushion measuring upwards of 6% at some FLs...)

Yes, exactly! The issue is that sharpness has become the single most important parameter by which people judge whether a lens is good or not. This is how many lenses have been sold - people read a review on the internet, stating that Lens A is very sharp, and they go and buy it. Sometimes, the producers will go for record breaking center sharpness values, leaving corners well behind, just to break another resolution record. I can name few lenses myself that follow that philosophy, which I personally do not fully approve (making lenses that review well, but are unexciting to actually use).

While there are several lenses that made people happy with having nearly zero distortion, other types of aberrations seems to be even less important for the mass user. Having heard anyone mentioning that Lens X has a maximum value for astigmatism of 4% that Coma/RSA is almost perfectly well corrected, etc? Just few people are usually bothered by these things...

But I assume that's a bit of an off-topic.

By all measures, 8% distortion is quite a lot!
#13
I do agree (with the producers) that most flaws can be fixed to an extent, but the lack of sharpness can not. Besides, most people are probably using these lenses with the default corrections on, and not going to extra lengths to disable them. I know I wouldn't probably care a lot, if only because I'm always shooting JPEG anyway.
#14
"Corrections on" means loss of sharpness/resolution.
#15
(01-03-2020, 05:58 PM)Brightcolours Wrote: "Corrections on" means loss of sharpness/resolution.

Only in JPEG, where the engine will also apply diffraction compensation and sharpening. For people who are shooting JPEG, this means that you get an usable image, ready to be sent right away to the client; or that you can have nice looking photos from your vacation that you can look on your TV by just connecting the camera.
#16
Only in JPEG? What?

If you shoot RAW, you still have to apply corrections. And lose the same sharpness/resolution you lost in the camera's JPEG engine.
#17
(01-04-2020, 12:22 PM)Brightcolours Wrote: Only in JPEG? What?

If you shoot RAW, you still have to apply corrections. And lose the same sharpness/resolution you lost in the camera's JPEG engine.

There's no switch "Corrections on" in any RAW convertor that I have used. Maybe in some pedestrian ones, there is such thing, but most of them allows you to selectively apply individual corrections, and to a certain degree. And even if you have enabled them in your camera, your RAW convertor might decide to apply different corrections on its own. And this will not modify the RAW file, but the rendered JPEG/TIFF image that you export based on the modifications that you have made to the image. This is why I believe your statement is not fully accurate without additional context.

If you apply CA correction with moderation, the actual result might produce sharper-looking image as CA corrections in software have little detrimental effect and the outcome of those being tamed results in positive net effect.

I would bet that, if this is not already the case, computational photography will be able to deal with lens' distortion with lower loss of corner detail.
#18
(01-04-2020, 01:55 PM)faint Wrote: There's no switch "Corrections on" in any RAW convertor that I have used. Maybe in some pedestrian ones, there is such thing, but most of them allows you to selectively apply individual corrections, and to a certain degree. And even if you have enabled them in your camera, your RAW convertor might decide to apply different corrections on its own. And this will not modify the RAW file, but the rendered JPEG/TIFF image that you export based on the modifications that you have made to the image.

That's not correct. Almost all RAW converters automatically apply more or less the same corrections if it is part of the system philosophy/manufacturers decision to rely on software correction. Only a few of the converters offer an option to switch these corrections off completely.

In any case: of course the correction is not applied to the RAW file, but on the result you have a look at/print/share online/give to others. Software correction of distortion always comes at a price (lower resolution), no matter if it is done in-camera or in the raw converter. The algorithms might be different, the results may vary between several products and compared to the in-camera conversion, however the loss of resolution is an unavoidable fact. The higher the distortion, the higher the loss of resolution.

That's why I prefer correction by glass over correction by software.

CA correction is a different story, because it's a fairly easy correction and yes, usually has a positive effect on the perceived sharpness (except for excessive CAs).
Editor
opticallimits.com

#19
(01-04-2020, 01:55 PM)faint Wrote:
(01-04-2020, 12:22 PM)Brightcolours Wrote: Only in JPEG? What?

If you shoot RAW, you still have to apply corrections. And lose the same sharpness/resolution you lost in the camera's JPEG engine.

There's no switch "Corrections on" in any RAW convertor that I have used. Maybe in some pedestrian ones, there is such thing, but most of them allows you to selectively apply individual corrections, and to a certain degree. And even if you have enabled them in your camera, your RAW convertor might decide to apply different corrections on its own. And this will not modify the RAW file, but the rendered JPEG/TIFF image that you export based on the modifications that you have made to the image. This is why I believe your statement is not fully accurate without additional context.

If you apply CA correction with moderation, the actual result might produce sharper-looking image as CA corrections in software have little detrimental effect and the outcome of those being tamed results in positive net effect.

I would bet that, if this is not already the case, computational photography will be able to deal with lens' distortion with lower loss of corner detail.

Confused about your point. 
This Tamron has a LOT of barrel distortion. You seem claim that less correction by the optics leads to sharper results. 
Correcting the barrel distortion of under-corrected lenses leads to..... less resolution. Whether the correction takes place in camera (JPEG) or in later RAW conversions is irrelevant.

CA leads to less sharp results, because different wave lengths lead to different sized projections. Correcting (not masking) CA leads to sharper results, because you map the different colours back to the same size.
#20
The folks at EPZ have reviewed the 24 and 35 Tamrons, hot on the heels of the 20 review. In a nutshell: both are very sharp, the 35 has little distortion and CA whereas the 24 has moderate CA that increase with stopping down and the (native) distortion on that one is ugly - about 6.1% (not as heinous as on the 20 but still rather steep).
  


Forum Jump:


Users browsing this thread:
1 Guest(s)