Olympus 8mm FE goes to Utah

[ArizonaMike]

I recently purchased the Olympus 8mm fisheye lens to use as a landscape pano lens so I could stop stitching photos together. As my screen name implies I live in the US South West and my wife and I travel to the US National Parks frequently and wanted a good lens for photographing the wide open desert regions. I had previously bought the Samyang 7.5mm for the same purpose, but found that I had difficulty properly framing my photos and hoped that the in-camera distortion correction of the Olympus lens would solve that problem. Our trip to Bryce Canyon and Zion National Parks were to be the test of that idea.

In preparation I set my E-M1.2 custom modes to reflect the 3 settings for the fisheye lens and took both the fisheye and the Olympus 12-50 on a hike into Bryce Canyon National Park. To avoid constantly changing lenses I basically used the 8mm on the hike down into the canyon and the 12-50 on the way back up, although at several places I took comparison images using both lenses as well as the Samyang 7.5 since I wanted to see if there were any color and quality differences. I learned quite a bit about using this lens, what to avoid and what to watch out for, and I will give some summary information at the end of this post. I first have to apologize for the length of the post as it is full of images as well as information regarding what I learned.

First, a couple of comments. These are the jpg images, automatically corrected in-camera by the E-M1.2, and I have not adjusted them so that people could see what the camera actually produces. They were run through Dxo's PhotoLab, but with the No Corrections preset so they have not been "fixed". You will see some photos with a bit of lens flare sun spots, which indicates how careful you have to be on a bright sunny day with a fisheye lens. The lens hood for this lens is (necessarily) small. Some of the horizons appear a bit off, but what appears to be the horizon often is not, in fact, level. No vibrancy or saturation has been added, except by the camera itself, and the raws are actually a bit more colorful.

Here are the 8mm images. I could not upload all of them (there is a limit to the number of images that can be included in a post) so all of these are 8mm and the 12-50mm images will have to be put into another post.

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

Summary:

1. Using the 8mm lens this way basically is the same as having a 3 position UWA zoom. The 3 settings are something like 5.5mm, 7.5mm and 10.5mm, and I believe that this lens, used this way, can easily replace my 9-18mm zoom.

2. In my view setting 1 on this lens (approximately 5.5mm) is not particularly useful. Things just appear too far away to make a decent image, at least for these photos.

3. There is no exif information in the images to indicate what auto setting (1, 2 or 3) was used when taking the images, so it is very difficult to determine what setting was used with any particular photo. I found I used the Narrow setting (3) for about 60% of the images, the Middle setting (2) for about 30% and the Wide (1) for about 10%. I do not expect to use the Wide setting in the future.

4. The jpgs turned out to be much more colorful than I expected, based on the images in the EVF. In fact I worried during the hike that the lens would not produce the colors of the canyon as they appeared washed out in the EVF but, as you can see, they are very colorful, and do reflect the colors in the canyon accurately.

5. The 8mm lens is not a macro lens, but you can focus very close to an object, and I am quite happy with the flower images.

6. I took a fair number of HDR images with the FE lens, but apparently if you use the HDR button to set the HDR settings with this lens you do not get the auto correction and my HDR shots all will have to be manually de-fished.

These shots are all from Bryce Canyon and I will have to post the Zion shots in a separate post if anyone is interested, as with the 12-50mm shots. I did not see any color or IQ differences between the lenses.

I would like to see comments.

 

[wjiang]

I'm curious about replacing your 9-18mm - you lose a lot of resolution when reprojecting, and the output still isn't actual anything close to rectilinear. I'm guessing you don't actually need a rectilinear UWA then?

 

[ArizonaMike]

I'm curious about replacing your 9-18mm - you lose a lot of resolution when reprojecting, and the output still isn't actual anything close to rectilinear.

The 9-18 has been a good lens for me, but the real question was whether or not it was wide enough for the photos I want to take. Mostly it is, but in places like the US National Parks, where the vistas are very wide, I end up stitching with the 9-18, so I wanted something a bit wider.

I thought long and hard about the 7-14 which would have been wide enough, but it is a fairly heavy lens and I did not want to carry any more weight than I needed to on long hikes in the desert heat. The sun in Arizona and Utah is very, very strong and hiking in the open (there really is very little cover to provide shade) just drains all of my energy, so I want the least weight I can get away with and that is why I am now shooting with the Olympus rather than the Canon FF I used to have.

My photos are generally only for family use, and only seen by family. You are right when you said that the auto correction of the FE is not truly rectilinear, but it is close enough so that the difference is not noticeable to me as long as I am taking landscape photos. In fact I spent some time comparing the Olympus auto correction jpgs with the output from both Dxo's PhotoLab and Lightroom on the raws, which is fairly rectilinear, and decided that I actually preferred the Olympus corrections. The PL and LR corrections were very good as long as I did not want anything wider than about 9mm, but if that is the only width that I needed I could have just continued using my 9-18.

I'm guessing you don't actually need a rectilinear UWA then?

I think that is an accurate statement.

If I find I do want or need a more rectilinear image I can always run the raws through PL or LR or PTLens. For me the compromise between the convenience of having images that are "almost" rectilinear and not having to manually crop each UWA image seems worth it. I guess I will see if it still is in 6 or 9 months. For the moment I am pretty well satisfied with what I get automatically from the Medium and Narrow settings on the fisheye.

 

[ArizonaMike]

@wjiang

Your post made me stop and think about how I am processing these images and the short-cut I took in accepting the auto-corrected jpgs instead of working with the raws.

I had first (and tentatively) decided to see how things worked if I just took the output jpgs, with the in-camera auto correction based on the setting that I used, and processed them instead of the raws, and that is what I did with the images I posted. In fact they were not post processed at all, other than to remove the exif data, and they seemed quite good to me.

However, after responding to your question ("I'm guessing you don't actually need a rectilinear UWA then?"), I decided to try this whole thing using the raws instead and processing them two different ways. First, using the Olympus auto correction in the Olympus Viewer and second using PhotoLab.

I ran the raws through the Olympus Viewer, batch cropping them using the Olympus fisheye correction, and then processed the resulting tiffs. Next I loaded the raws into PhotoLab. I had created 3 crop presets in PhotoLab when I first got the lens, each corresponding to one of the 3 Olympus settings, and went through the raws using those presets based on what I thought looked best for the shot, and processed those. The comparisons between the 3 sets of images is interesting, and I am rethinking my tentative decision to use the jpgs instead of the raws.

I just wanted to thank you for posting the question that made me re-think my fisheye strategy.

 

[BosseBe]

@wjiang

Your post made me stop and think about how I am processing these images and the short-cut I took in accepting the auto-corrected jpgs instead of working with the raws.

I had first (and tentatively) decided to see how things worked if I just took the output jpgs, with the in-camera auto correction based on the setting that I used, and processed them instead of the raws, and that is what I did with the images I posted. In fact they were not post processed at all, other than to remove the exif data, and they seemed quite good to me.

However, after responding to your question ("I'm guessing you don't actually need a rectilinear UWA then?"), I decided to try this whole thing using the raws instead and processing them two different ways. First, using the Olympus auto correction in the Olympus Viewer and second using PhotoLab.

I ran the raws through the Olympus Viewer, batch cropping them using the Olympus fisheye correction, and then processed the resulting tiffs. Next I loaded the raws into PhotoLab. I had created 3 crop presets in PhotoLab when I first got the lens, each corresponding to one of the 3 Olympus settings, and went through the raws using those presets based on what I thought looked best for the shot, and processed those. The comparisons between the 3 sets of images is interesting, and I am rethinking my tentative decision to use the jpgs instead of the raws.

I just wanted to thank you for posting the question that made me re-think my fisheye strategy.

Interesting, could you post some pictures for comparison? (Between Oly correction and PL correction).

 

[ArizonaMike]

I suppose there really are 3 images to compare - the auto corrected jpg, the Olympus Viewer 3 fisheye correction (which is not the same as the in-camera correction) the the PL/LR corrections. I have now tried all 3, and will upload samples of all 3 corrections, but the Olympus Viewer correction is automatic and does not allow the user to select where to place the crop, so it can not be used well for composition unless the original image was composed with the relevant part in the middle of the image. One other thing to consider when looking at these images is that while the PL and in-camera crops were manually selected the Olympus Viewer crop is not, so the in-camera and PL crops may include more or less of the image width, depending upon how it was taken and cropped. I will try to include only those where the 3 images are largely the same subject material.

OOC jpg auto corrected:

Subscribe to see EXIF info for this image (if available)

Olympus Viewer 3:

Subscribe to see EXIF info for this image (if available)

PhotoLab:

Subscribe to see EXIF info for this image (if available)

I found the yellow flowers on the right side of this photo to be a bit too "stretched" by PhotoLab and thought that the in camera correction was better, but it is a matter of taste, I suppose.
-----------

OOC jpg auto corrected:

Subscribe to see EXIF info for this image (if available)

Olympus Viewer:

Subscribe to see EXIF info for this image (if available)

PhotoLab:

Subscribe to see EXIF info for this image (if available)

These images seem to me to show the problem with using the PhotoLab distortion correction. The PL image is very different in perspective than the other two and, as I remember, different from the way the formation actually looked. It seems to me that the in-camera and OV results are more like the original image.
--------------
OOC jpg auto corrected:

Subscribe to see EXIF info for this image (if available)

Olympus Viewer:

Subscribe to see EXIF info for this image (if available)

PhotoLab:

Subscribe to see EXIF info for this image (if available)

Again, here, the PhotoLab image correction is very different from the in-camera and Olympus Viewer corrections, and not as close to what it originally looked like, as best I remember.
--------------
OOC jpg auto corrected:

Subscribe to see EXIF info for this image (if available)

Olympus Viewer:

Subscribe to see EXIF info for this image (if available)

PhotoLab:

Subscribe to see EXIF info for this image (if available)

These look pretty much the same for all 3 correction types. This shows something of the problem for me in deciding what correction to use. PhotoLab allows me to use the original raws, but it sometimes produces an image that I do not find either pleasing or realistic. The Olympus Viewer does a nice automatic correction, but it is time consuming (each raw image correction takes about 23 seconds, so 100 corrections takes more than 30 minutes) and it may then have to to be further cropped. The in-camera images seems the easiest to use, but it is a jpg, so there is only so much I can do with it.

If I did not like the lens so much I would just return it and look for an alternative, but I find the idea of a light weight UWA "zoom" appealing and I like the ability to be able to properly frame the image that I am taking using the in-camera correction.

COMMENT:

It is not easy to see the different perspectives when viewed on the forum, but if you download 3 images and look at them in rapid succession it is easy to see how they differ.

 

[BosseBe]

I suppose there really are 3 images to compare - the auto corrected jpg, the Olympus Viewer 3 fisheye correction (which is not the same as the in-camera correction) the the PL/LR corrections. I have now tried all 3, and will upload samples of all 3 corrections, but the Olympus Viewer correction is automatic and does not allow the user to select where to place the crop, so it can not be used well for composition unless the original image was composed with the relevant part in the middle of the image. One other thing to consider when looking at these images is that while the PL and in-camera crops were manually selected the Olympus Viewer crop is not, so the in-camera and PL crops may include more or less of the image width, depending upon how it was taken and cropped. I will try to include only those where the 3 images are largely the same subject material.

OOC jpg auto corrected:

View attachment 655005

Olympus Viewer 3:

View attachment 655006

PhotoLab:

View attachment 655007

I found the yellow flowers on the right side of this photo to be a bit too "stretched" by PhotoLab and thought that the in camera correction was better, but it is a matter of taste, I suppose.
-----------

OOC jpg auto corrected:

View attachment 655008

Olympus Viewer:

View attachment 655009

PhotoLab:

View attachment 655010

These images seem to me to show the problem with using the PhotoLab distortion correction. The PL image is very different in perspective than the other two and, as I remember, different from the way the formation actually looked. It seems to me that the in-camera and OV results are more like the original image.
--------------
OOC jpg auto corrected:

View attachment 655012

Olympus Viewer:

View attachment 655013

PhotoLab:

View attachment 655014

Again, here, the PhotoLab image correction is very different from the in-camera and Olympus Viewer corrections, and not as close to what it originally looked like, as best I remember.
--------------
OOC jpg auto corrected:

View attachment 655015

Olympus Viewer:

View attachment 655016

PhotoLab:

View attachment 655017

These look pretty much the same for all 3 correction types. This shows something of the problem for me in deciding what correction to use. PhotoLab allows me to use the original raws, but it sometimes produces an image that I do not find either pleasing or realistic. The Olympus Viewer does a nice automatic correction, but it is time consuming (each raw image correction takes about 23 seconds, so 100 corrections takes more than 30 minutes) and it may then have to to be further cropped. The in-camera images seems the easiest to use, but it is a jpg, so there is only so much I can do with it.

If I did not like the lens so much I would just return it and look for an alternative, but I find the idea of a light weight UWA "zoom" appealing and I like the ability to be able to properly frame the image that I am taking using the in-camera correction.

COMMENT:

It is not easy to see the different perspectives when viewed on the forum, but if you download 3 images and look at them in rapid succession it is easy to see how they differ.

I did as you said and there really is a clear difference, but what is correct is impossible for me to say as I don't have any real world reference.
There is a slight difference between OOC JPEG and OV versions, some more barrel correction in the OV version?
Between OV and PL I see a lot more difference, seems like a lot more barrel correction.
Since PL's corrections are measured (probably only on one sample of the lens), maybe your lens is better and need less correction.
Have you tried to changed the default corrections in PL?
Great pictures, BTW.

 

[ArizonaMike]

I did as you said and there really is a clear difference, but what is correct is impossible for me to say as I don't have any real world reference.

My wife and I travel to Bryce almost every year and we go for a hike through the canyons on almost every visit, but even with all of those visits it is nearly impossible for me to remember exactly what the formations are supposed to look like. That is what the photos are supposed to do for me.

Still, I do not think that the PL distortion correction, which is supposed to be the most accurate, is right in those photos. Perhaps the PL corrections work best if the camera and lens are level with the ground, and that is not always possible.

Since PL's corrections are measured (probably only on one sample of the lens), maybe your lens is better and need less correction.
Have you tried to changed the default corrections in PL?
Great pictures, BTW.

Interestingly enough my first thought was if I could somehow change the default settings for the PL corrections, but I do not know where the information is stored, or how I could change it if I did know. And, even if I knew where the correction data was stored, and how to change it, I would not know how to create a new set of corrections that would work well under all circumstances. However I will look into it as not knowing how to do something has never stopped me in the past from trying anyway ...

And thank you for the compliment. The colors may look strong, but that is actually what they look like in person.

 

[Growltiger]

Thank you for the comparison shots.

The first picture in the post above - the one with the long log at bottom right, is a example that shows why I'm sticking with rectilinear lenses. The Photolab version, which produces a true rectilinear image, shows a fairly straight log. The other two show an unnaturally curved log.

I appreciate there are cases where the eye doesn't notice the fisheye distortion, but I'm quite sensitive to it. I also appreciate that to many people ultrawide rectilinear distortion looks bad too, but I'm used to it and try to avoid putting various things at the edges, such as faces.

 

[wjiang]

With my 7.5mm FE I rarely use either the native fisheye projection or rectilinear on something so wide actually - usually something slightly less fishy in Hugin. I guess that's what you're finding too. It's nice to have an in-camera preview though.

 

[BosseBe]

I was at Bryce canyon last year, but stayed at the top, so the colours look right to me but the formations I have not seen.
I don't think you can change the defaults settings in Pl, but you can make your own setting to use (I believe there is a setting to choose what "profile" to use by default).

 

[ArizonaMike]

Thank you for the comparison shots.

The first picture in the post above - the one with the long log at bottom right, is a example that shows why I'm sticking with rectilinear lenses. The Photolab version, which produces a true rectilinear image, shows a fairly straight log. The other two show an unnaturally curved log.

I appreciate there are cases where the eye doesn't notice the fisheye distortion, but I'm quite sensitive to it. I also appreciate that to many people ultrawide rectilinear distortion looks bad too, but I'm used to it and try to avoid putting various things at the edges, such as faces.

It is probably still a bit early for me to be making generalizations since I have only just started really using the lens, but what I am beginning to think is that the PhotoLab corrections (and the LR corrections as well, since they seem to be largely identical) likely produce the most accurate images provided the camera/lens combination is horizontal when the image is taken and that the image is cropped to no more than perhaps 7.5-8mm equivalent. The big differences I have seen between the Olympus and PL corrections have all been at the edges of the image (top, bottom or sides) or when the camera is not level.

Going through the raw images using PL and applying either the "Narrow" or "Medium" crops as presets usually yields a very good image, more rectilinear than that from the jpgs, but at the edges of the images the in-camera corrections are better. I am just not sure what that means for my photo trips and suspect I will not know until I have processed more images. I still have the Zion National Park stuff to go through as well.

By the way, I had not noticed the differences in the log between the 3 images until you pointed it out. You are, of course, correct. One thing I did notice is that the FE lens was all I needed in the canyons. I took, and used, the 12-50 mostly for color and IQ comparisons, but only needed it for one shot when I was trying to get the photo of a bird nesting in the rock formations.

Given this uncertainty perhaps I will hold on to the 9-18 until I see how all of this shakes out.

 

[ArizonaMike]

With my 7.5mm FE I rarely use either the native fisheye projection or rectilinear on something so wide actually - usually something slightly less fishy in Hugin. I guess that's what you're finding too. It's nice to have an in-camera preview though.

I went through your tutorial on using Hugin (thank you for that), but I have never been able to get a really good correction using it. I suspect that I am just not familiar enough with Hugin to know how to use it, but I always had problems with that part of the image laying along a horizontal line in the middle of the image. If it was curved up or down in the middle I could not find a way to correct it.

I probably need more experience with it, and now I have a fair number of images to play around with. For me this is still a work in progress.

 

[wjiang]

I suspect that I am just not familiar enough with Hugin to know how to use it, but I always had problems with that part of the image laying along a horizontal line in the middle of the image. If it was curved up or down in the middle I could not find a way to correct it.

It's all to do with how the different projections work mathematically. Cylindrical-like projections (cylindrical, Panini) will straighten verticals but horizontals will be curved except for the one right in the centre. You can use the pitch adjustment in Hugin to get that centre where you want it.