Pages: [1]   Go Down

Author Topic: Canon 17mm T/S: reducing fringing around windows  (Read 1378 times)

mshea

  • Full Member
  • ***
  • Offline Offline
  • Posts: 244
Canon 17mm T/S: reducing fringing around windows
« on: January 13, 2019, 01:26:51 pm »

I often run up against the problem of fringing at window edges. This will happen whether I'm shooting with midday light or at twilight. The same fringing happens with bright exterior signage around twilight. Brightly lit or dark interiors, the problem's pretty much the same.

I usually use 5 - 6 stops of underexposure for my interior HDR photos and I normally use LR HDR.

Here's a cropped example with 5 stops underexposure. It was a quick test shot. The exterior light was too bright and I could easily have added two more stops of underexposure, so the window highlights were blown out (reduced highlights in LR 100%). But it illustrates the problem.

I've occasionally used the Dehaze brush, but I've found it to be a somewhat crude tool for this purpose.

I'm assuming this is an inherent characteristic of the 17 T/S. Or might the camera sensor (A7RIII) also be contributing to the problem? I'm wondering if there might be another workaround. And does the T/S 24 V.II perform better?

Curious to know others' experiences with this lens.

Thanks,
Merrill
Logged

David Eichler

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 823
    • San Francisco Architectural and Interior Photographer
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #1 on: January 13, 2019, 05:27:06 pm »

Fringing is the wrong term for this artifact. It is not the result of the lens. It is an electronic/processing phenomenon. Your HDR image is reflecting the sensor overload affecting the areas around the windows in the lighter exposures.
Logged

Slobodan Blagojevic

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 18090
  • When everyone thinks the same, nobody thinks
    • My website
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #2 on: January 13, 2019, 06:01:32 pm »

Fringing is the wrong term for this artifact. It is not the result of the lens. It is an electronic/processing phenomenon. Your HDR image is reflecting the sensor overload affecting the areas around the windows in the lighter exposures.

Blooming?

LesPalenik

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5339
    • advantica blog
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #3 on: January 14, 2019, 04:28:36 am »

Whatever the name of this phenomenon is, Adobe Lightroom (most likely also Camera Raw) usually cleans it up quite nicely (in the Lens Correction, Color, Remove Chromatic Aberration). 
Logged

alan_b

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 318
    • West Coast Architecture + Interiors Photographer
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #4 on: January 14, 2019, 12:16:17 pm »

It's flare. 

You need to reduce the abrupt brightness differences on site.  Fill light works, or hold up a dodging mask in front of the lens to block the brightest areas from the lens, then composite later.
Logged

David Eichler

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 823
    • San Francisco Architectural and Interior Photographer
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #5 on: January 14, 2019, 02:05:54 pm »

It's flare. 

You need to reduce the abrupt brightness differences on site.  Fill light works, or hold up a dodging mask in front of the lens to block the brightest areas from the lens, then composite later.

While some flare will often occur in situations such as the one illustrated, even with the most flare-resistant lenses, the predominant effect seen here is due to sensor overload that occurs around very-high-contrast edges and causes the very bright areas to bleed over into the adjacent darker areas, which can look similar to flare. However, flare is an optical phenomenon and sensor overload is an electronic one. And, yes, supplemental fill lighting is one common way to deal with situations such as the one in the example, although this is not practical with all subjects.
Logged

JaapD

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 303
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #6 on: January 15, 2019, 01:23:53 am »

I don’t think its sensor blooming. Modern sensor designs have an anti-blooming functionality, capable of about 150x over saturation, or more. All excess charge then flows to the substrate.

In case the sensor is blooming then it is only blooming to its neighbor pixel, definitely not over a wide area.

I definitely think its flare, sometimes called ‘low frequency drop’. It’s purely an optical phenomenon. In pictures like these it’s rapidly visible even with the best flare resistant optics. Only 1% of flare caused by the bright areas will result in lightening up of the dark areas.

Regards,
Jaap.

« Last Edit: January 15, 2019, 01:43:42 am by JaapD »
Logged

David Eichler

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 823
    • San Francisco Architectural and Interior Photographer
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #7 on: January 15, 2019, 03:35:15 am »

I don’t think its sensor blooming. Modern sensor designs have an anti-blooming functionality, capable of about 150x over saturation, or more. All excess charge then flows to the substrate.
Regards,
Jaap.

Can you provide an authoritative source to back up this assertion? Flare should be evident regardless of the level of exposure. I am willing to bet that the darker component exposures of this HDR image do not exhibit the hazy bleeding into the neighboring shadow areas seen in the example, at least not nearly to the same degree as in the example HDR, and that hazy bleeding over becomes progressively stronger in the overexposed component images, as the overexposure becomes greater.

This is a description of the phenomenon to which I refer: http://www.optique-ingenieur.org/en/courses/OPI_ang_M05_C06/co/Contenu_14.html

Perhaps you have been confused by comparisons of CMOS sensors to CCD sensors? I have seen such comparisons claim that CMOS sensors do not exhibit "blooming" when overloaded, but I think what they really mean is that CMOS sensors exhibit much less of this than CCD sensors under typical conditions. However, the lighting conditions shown in the HDR example image reflect a pretty extreme dynamic range.

Some additional information: https://graphics.stanford.edu/courses/cs178/lectures/sensors-24apr14.pdf

And there is this: https://www.fujifilmusa.com/support/ServiceSupportProductContent.do?dbid=881329&prodcat=879660&sscucatid=664262

« Last Edit: January 15, 2019, 04:23:16 am by David Eichler »
Logged

Martin Kristiansen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1527
    • Martin Kristiansen
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #8 on: January 15, 2019, 04:10:14 am »

Can I also have a go?

The flare around the windows I saw perfectly natural and is not a fault on lens, sensor or software. It’s a failure to see. The bright exposure used to give detail in the nterior records the flare around the window because it’s closest to the light source and is the next brightest thing after the windows themselves. When looking at this frame in the brighter area around the window looks natural. The you replace the overexposed windows with underexposed frames to hold some detail in the windows but the very bright areas immediately around the windows is not replaced. And is a left over from the over exposed frame used for the interior. Now it looks odd because there is no logical source we can see that explains this over bright anomaly.

We assume that replacing the windows will sort out our problem but the edge of the window is not the cut off area for the overexposed light from outside. That cut off area is a gradient from the window edge extending inward and as such is very difficult to deal with.

My experience with this is to not tone down the outside area so much. Allow it to be brighter. It will look more natural. Then don’t work with centre weighted exposure inside. Allow it to be darker. Yes that will give a bit of noise but modern sensors and noise software will deal with it easily. Brighten this exposure as much as is needed in post. What this means is the difference between the exposures is not as great. That will mean much less work to fix the flare afterwards if it even shows up.

Anyway that not just my theory it’s also my approach and my clients have not complained.
Logged
Commercial photography is 10% inspiration and 90% moving furniture around.

David Eichler

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 823
    • San Francisco Architectural and Interior Photographer
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #9 on: January 15, 2019, 04:18:51 am »

Can I also have a go?

The flare around the windows I saw perfectly natural and is not a fault on lens, sensor or software. It’s a failure to see. The bright exposure used to give detail in the nterior records the flare around the window because it’s closest to the light source and is the next brightest thing after the windows themselves. When looking at this frame in the brighter area around the window looks natural. The you replace the overexposed windows with underexposed frames to hold some detail in the windows but the very bright areas immediately around the windows is not replaced. And is a left over from the over exposed frame used for the interior. Now it looks odd because there is no logical source we can see that explains this over bright anomaly.

We assume that replacing the windows will sort out our problem but the edge of the window is not the cut off area for the overexposed light from outside. That cut off area is a gradient from the window edge extending inward and as such is very difficult to deal with.

My experience with this is to not tone down the outside area so much. Allow it to be brighter. It will look more natural. Then don’t work with centre weighted exposure inside. Allow it to be darker. Yes that will give a bit of noise but modern sensors and noise software will deal with it easily. Brighten this exposure as much as is needed in post. What this means is the difference between the exposures is not as great. That will mean much less work to fix the flare afterwards if it even shows up.

Anyway that not just my theory it’s also my approach and my clients have not complained.

If what you are trying to say were so, the darker areas immediately around the windows would have good contrast. However, the example exhibits a hazy, low-contrast appearance in those areas, a result of sensor overload ("bloom"), possibly with some flare as well (the darker exposures would reveal whether there is any flare component).

Also, your use of the term flare is inaccurate. In photographic terminology, flare is an optical phenomenon that results from the lens or reflections inside the camera, or some combination of these.
« Last Edit: January 15, 2019, 04:28:08 am by David Eichler »
Logged

Martin Kristiansen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1527
    • Martin Kristiansen
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #10 on: January 15, 2019, 04:35:37 am »

If what you are trying to say were so, the darker areas immediately around the windows would have good contrast. However, the example exhibits a hazy, low-contrast appearance in those areas, a result of sensor overload ("bloom"), possibly with some flare as well (the darker exposures would reveal whether there is any flare component).

Also, your use of the term flare is inaccurate. In photographic terminology, flare is an optical phenomenon that results from the lens or reflections inside the camera, or some combination of these.

Yes yes of course. But you want to fix it or debate semantics? Actually I know the answer to that question.
Logged
Commercial photography is 10% inspiration and 90% moving furniture around.

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8913
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #11 on: January 15, 2019, 05:44:11 am »

I often run up against the problem of fringing at window edges. This will happen whether I'm shooting with midday light or at twilight. The same fringing happens with bright exterior signage around twilight. Brightly lit or dark interiors, the problem's pretty much the same.

I usually use 5 - 6 stops of underexposure for my interior HDR photos and I normally use LR HDR.

Hi Merrill,

It's not immediately clear to me whether you take multiple exposures, or just 2 (with 1 underexposed for the highlights).
When I do exposure bracketing, I take a series of images with no more than 1.3 stops intervals. That allows (in my case with SNS-HDR) to produce very natural looking tonemappings of very high dynamic range scenes.

Without seeing the individual exposures, it's hard to judge whether the edge issues arise from :
a. The lens rendering
b. Sensor bloom
c. A Raw converter issue
d. Tonemapping issues

Cheers,
Bart
« Last Edit: January 15, 2019, 08:12:24 am by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==

Aram Hăvărneanu

  • Full Member
  • ***
  • Offline Offline
  • Posts: 214
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #12 on: January 15, 2019, 06:08:00 am »

Bart is right, we need more information.

However, I think it's a tonemapping issue.
Logged

32BT

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3095
    • Pictures
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #13 on: January 15, 2019, 06:28:03 am »

1. The problems around the chains definitely indicate that the exposure interval in the bracket is too great. That would be solved most likely by more exposures with smaller step.

2. The glare near the windows could be a result of flare between lens elements. You might want to check if your bracketing is honnoring a single aperture. It also can be solved by shooting more exposures with smaller interval as that allows you to selectively blend the desired images.

3. In my experience, if you have to apply highlight recovery, the bracket might be considered a fail as these types of problems, especially around the chains, will show up. Having said that: it's been a while since doing a true hdr, and I have no experience with that lens.

Another $0.02 down the drain...
Logged
Regards,
~ O ~
If you can stomach it: pictures

David Eichler

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 823
    • San Francisco Architectural and Interior Photographer
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #14 on: January 15, 2019, 08:40:59 pm »

1. The problems around the chains definitely indicate that the exposure interval in the bracket is too great. That would be solved most likely by more exposures with smaller step.

2. The glare near the windows could be a result of flare between lens elements. You might want to check if your bracketing is honnoring a single aperture. It also can be solved by shooting more exposures with smaller interval as that allows you to selectively blend the desired images.

3. In my experience, if you have to apply highlight recovery, the bracket might be considered a fail as these types of problems, especially around the chains, will show up. Having said that: it's been a while since doing a true hdr, and I have no experience with that lens.

Another $0.02 down the drain...

My professional specialty is architectural and interiors photography. You will find examples illustrating my capabilities at the website url below. I frequently deal with the sort of technical challenge illustrated in the OP's example. I am conversant with HDR and have used various HDR programs over the years in the course of my work. I use the Canon 17mm ts-e frequently in the course of my work. Therefore, I think it is fair to say that I have some reasonable expertise in this subject, though I hardly claim to know everything about it.

As is typical of LuLa, many of the people commenting in this discussion do not show examples to demonstrate their capabilities or they hide behind pseudonyms, or both, which does not give me much confidence in their credibility.

Based on my experience, neither tonemapping nor the size of the exposure intervals (assuming typical intervals of 1 to 2 stops between exposures) is the cause of the artifact. It will often show up in any automated process for combining varied exposures to subdue an extreme dynamic range when the "blooming" is evident in the majority of the lighter and middle component exposures, though variations in the tonemapping may make the effect more or less evident.

With Lightroom's HDR, I have sometimes been able to subdue this effect with the LR adjustment brush, but this is not a panacea. Another option sometimes is to retouch in Photoshop. However, the techniques I most often use to deal with this technical challenge involve using supplementary lighting, with or without HDR (mostly without).

 
« Last Edit: January 16, 2019, 01:34:24 am by David Eichler »
Logged

mshea

  • Full Member
  • ***
  • Offline Offline
  • Posts: 244
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #15 on: January 16, 2019, 12:50:28 pm »

Well, all sorts of opinions here! To further explain, I established a base-line exposure, in which there was some shadow detail, even though the window highlights were obviously blown out. I then decreased the exposures at one-stop intervals for a total of five exposures. The aperture was a constant f/11.  I'm not certain, but I might have used center-weighted metering. I'll certainly go with Multi at the next shoot.

It's obvious I could have included a few more stops of underexposure (I should have zoomed in on the LCD to check the window highlights during shooting). This was just a test, so maybe that will improve things next time around. I'm volunteering my time to shoot this place, so I'm trying to avoid having to do a whole interior lighting setup.

Thanks,
Merrill



Logged

David Eichler

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 823
    • San Francisco Architectural and Interior Photographer
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #16 on: January 16, 2019, 01:24:38 pm »

Well, all sorts of opinions here! To further explain, I established a base-line exposure, in which there was some shadow detail, even though the window highlights were obviously blown out. I then decreased the exposures at one-stop intervals for a total of five exposures. The aperture was a constant f/11.  I'm not certain, but I might have used center-weighted metering. I'll certainly go with Multi at the next shoot.

It's obvious I could have included a few more stops of underexposure (I should have zoomed in on the LCD to check the window highlights during shooting). This was just a test, so maybe that will improve things next time around. I'm volunteering my time to shoot this place, so I'm trying to avoid having to do a whole interior lighting setup.

Thanks,
Merrill

If you are going to meter the scene, I would suggest using a spot meter to read the lightest and darkest areas in the scene. Or, you could dispense with the meter and use the histogram. I am not sure you needed any darker exposures for this example. You don't necessarily need an elaborate lighting set up just to hold detail in the windows if the quality of light for the interior is good. A couple of strobes would suffice. But you would need to have good Photoshop and lighting skills. Or, you could shoot at twilight, when the light level in the windows is balanced with that of the interior.
Logged

32BT

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3095
    • Pictures
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #17 on: January 16, 2019, 01:50:56 pm »

Would it perhaps be an alternative option to stack a whole lot of dark exposures into a single 32bit float image where you have more shadow leeway, but none of thre downsides of overexposure eating into high-contrast edges?
Logged
Regards,
~ O ~
If you can stomach it: pictures

mshea

  • Full Member
  • ***
  • Offline Offline
  • Posts: 244
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #18 on: January 16, 2019, 03:30:01 pm »

Yes, twilight is definitely the time to shoot. I was shooting at least an hour earlier than I would have preferred. Trouble is, the place closes at 4:00.

Merrill
Logged

David Eichler

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 823
    • San Francisco Architectural and Interior Photographer
Re: Canon 17mm T/S: reducing fringing around windows
« Reply #19 on: January 16, 2019, 04:02:05 pm »

Would it perhaps be an alternative option to stack a whole lot of dark exposures into a single 32bit float image where you have more shadow leeway, but none of thre downsides of overexposure eating into high-contrast edges?

I don't see how that could do anything for you in this regard. After all, darker exposures will have less shadow detail, not more. You need the lighter exposures for the shadow detail. Whenever you have an extreme dynamic range and extreme, abrupt transitions from light to dark, you are going to have this problem, whether from blooming or flare, or both.

Btw, while not immune from flare, the Canon 17mm ts-e lens has very good to excellent flare resistance, especially for a lens of such complex design.
Logged
Pages: [1]   Go Up