Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
5D MK IV is out officially
#31
A serious question to you, JoJu, since I keep on wondering about it:

Is it doable to use the f1.8 APS-C zooms from Sigma on the 1.3x crop APS-H Sigma body without encountering problematic vignetting?

 

On the panorama.. Which ISO did you shoot at? I clicked on the link and looks at the left side building 100%, and immediately I noticed pretty strong to VERY strong magenta and green noise on the "white" building panels. Going down to the roll doors, it becomes green and maganeta blotches one usually sees with very high ISO and bad NR.

Going down to the bricks in the street pavement, it is green and magenta big patches all over the place. Some photos you stitched seem to have very strong NR  leading to the weak magenta and green patches, others just have very strong green and magenta patches. Not sure why this image has you impressed with the DP2 Merril?

 I'd have to use ISO 3200 with my EOS M ($200 at the time) to get such bad IQ, I think.

 

Do I remember it wrong, or does the DP2 Merril use the old every-pixel-reads-separete-RGB-value sensor (the new Quattro ones do not)?

#32
When I had the 18-35/1.8, I also put it just for fun in front of the D800 without activated DX-format switch. I remember there was plenty of distance to the visibly darker image-circle areas. But I didn't test or shoot pictures at that situation, not to mention measuring the vignetting. I guess there is some more vignetting wide open at 18-20 mm - but which other wide angle has no? I could still send the 20/1.4 Art FF in to be on the very safe (and big and heavy and chunky) side of things. I just guess (again) after stopping down 2 or three stops, it will become hard to see vignetting.

 

With the panorama were two apps involved: Iridient, becasue it's much faster and allows more control including cropping than Sigma PhotoPro. And Affinity Photo. I recall to have been at the end of the panorama in the "develop persona" (meaning, the raw-developing module of the app) to lift up the shadows. I cannot first could not, but now I do see what you wrote about, but that doesn' mean much as there is no wide gammut monitor involved on my side. You're right. I suspect, the shadows were underexposed already and careless denoising did the rest. I just don't feel much at home with Irdient's or Affinty Photo's noise treatment. Maybe the lifting up of the shadows in Affinity brought in the effect. It was clean when I looked at it before Iridient's batch processing. Ok, "clean enough" for me who doesn't bother much about noise anyway. And honestly, since Sigma's PP 6.3 is supersloooooow in displaying the results, I just leave the noise part on standard settings.

 

Btw. the magenta tint could also Be caused by the light temperature, the sky was very much evening magenta and there are walls reflecting that to the side of the building. When I look at the upper left side, I also see CA which I didn't correct before.

 

But noise at 100 ISO (to answer this part of your question as well) can be improved. It was the sharpness which keeps impressing me. From a comparatively cheap (yet old) camera.
#33
Ok, thanks for the info, it is interesting that the image circle may be big enough to cover APS-H!

Wide open "vignetting" is normal light fall off, which is not really the same as black corners from a too small image circle where it counts.

#34
I will look tonight if I still have one of those pictures made with the 18-35 on FF, but I'm afraid to have them deleted. I recall the distance between each DX-corner and FF corner was less than 50% dark, but I can't say much about IQ of this zone. I know, it's a very exotic camera and for sure not the smallest mirrorless, I also know if the Fuji makes me happy for most purposes, the wish to get a new Sigma toy will decrease as I only would use it for maybe 1/10 of my pictures. The lack of a good, quick enough converter becomes a big deal as well.

 

Anyway, the APS-H version so far is only announced. These are neither available nor with a known price.

 

The reason for the 80 MP quick and dirty panorama was less the Sigma, than how much of the newly added 12 GB more RAM will be used to stitch this 12 TIFF pictures.

 

Thanks this time for your critical and correct observation. I will try to make a better version, just to be able to handle it better next time with a nice panorama situation. I know now I have to deal with the shadow noise before stitching.

 

Edit:

[Image: vignetting.gif]

 

I was looking a bit more into the differences. The vignetting charts from Sigma appear to be really concerning, from 100% down to 35% at the diagonal edge of Sigma's APS-C (diagonal roughly 28 mm). However, the APS-H (26.7×17.9 mm) diagonal is only 2 mm/radius more (32 mm).

 

So, the 20/1.4 appears to be a much better choice in terms of vignetting:

[Image: vignetting.gif]

 

But practically: Who makes reproductions at 18 mm wide open? If I use this lens at the sd quattro, it's for sure not for available light photography primarily, but because at normal apertures its' behaving like a very good prime, with the bonus of occassionally going artsy-fartsy and be able to focus in dark situations.

 

Comparatively: A Zeiss Batis 18/2.8 wide open has at 14 mm radius 60% vignetting, the 18-35 (at f/4!) 85% and the 20/1.4 (at f/2.8) more than 90%.

 

And I'm not sure what the infrared filter between mount and sensor will also eat up from this theoretical vignetting values.
#35
Quote:A serious question to you, JoJu, since I keep on wondering about it:

Is it doable to use the f1.8 APS-C zooms from Sigma on the 1.3x crop APS-H Sigma body without encountering problematic vignetting?
Let me answer (a part of) this question for you. I tested the 18-35 (not mine, 'fkorz) on my 1D Mark IV (the tinkerer that I am). At f/5.6 the vignetting is noticeable at the shortest focal length (no hard clipping though) but goes away the more you zoom in - it's nigh on invisible at 20mm. Wide open it is invisible at 28mm and 35mm and starts becoming progressively worse from 24mm on down. 18mm f/1.8 setting is unusable if you ask me, 20mm is marginal (both could probably be used for night shooting where the darkened corners would - with luck - blend into the darkness). Whether it depends (a lot) on the focusing distance, I don't know. It could be possible. I did get the degree of vignetting at 18mm f/5.6 to vary depending on focusing distance, but I didn't measure it scientifically. The (silly and half-hearted) test shots (originals with, well, full EXIF and whatnot) are here: https://cloud.mail.ru/public/WxXi/4zDPEou9S

 

Again: this was on Canon 1D4; whether the Sigma's coverage is any different is beyond my knowledge.

 

The 50-100 should be better in this regard, check out Dustin Abbott's review of that. I never even held it in my hands so I can't give you a first hand opinion. I would have loved to have that lens if I had been planning a total revamping of my kit.

#36
Canon's APS-H is a tiny bit larger than Sigma's, app. 1 mm per side more. Would be interesting, if Sigma themselves recommends that lens for the sd quattro H? Searched, but didn't find it yet.

#37
The dual pixel RAW capable of refocussing:

http://www.kamerabild.se/sites/kamerabil...k=eCcybuPD

http://www.kamerabild.se/sites/kamerabil...k=UTieoSLZ

from http://www.kamerabild.se/tester/vi-har-p...nodePage=3

#38
I have to admit that I am slightly puzzled how they do it - especially how they realize a certain range rather than just 2 possible settings.
#39
What I gather is that they can judge a distance from the shooting distance information, focal length and lens profile and the PD information. From that, within limits, they can reconstruct what is hidden in blurry areas with deconvolution.

 

A less sophisticated but kinda easy to understand example of deconvolution:

http://nl.mathworks.com/help/images/exam...rithm.html

http://www.deconvolve.net/bialith/Resear...ckblur.htm

 

With the phase difference data at hand, Canon can shift the focal plane. The normal DLO in DPP does not have that phase difference data to work with, and it only reconstructs unsharpness form the lens, AA-filter, and other aberrations. This advanced version is pretty neat?

#40
dpreview were less than overwhelmed by the feature. Yes you can adjust the focus point, a very little bit, but maybe not enough to actually be useful. For instance you couldn't correct for a shot that has focused on the nose rather than eye.

 

Also it appears to degrade image quality in the rest of the image.

  


Forum Jump:


Users browsing this thread:
2 Guest(s)