Panasonic Lumix S1R Conversion: An Exploration

[Andrea B.]

I made a few late fall botanical photos with the S1R before we get the Big Freeze. I've posted some Globemallow Vis and UV shots with conversion in Luminar. For the S1R, Luminar is much easier than trying to use SilkyPix. YMMV on that, of course.

 

Although a bit slow in rendering, Luminar has some nice basic tools along with a layering capability and tools to mask out or in any editing changes. I use the mostly the brush mask to add sharpening or details to parts of a flower. I don't know yet what are the best starting settings for the sharpening, details and denoising tools, but they all seem to work quite well. The white balance tool handles the S1R files very well also. I don't think that either the highlights or shadows tools work quite as well as those in Photo Ninja, but they are adequate. There are other tools I haven't tried yet. And there are a huge number of existing presets (called "looks") which might be useful to some. You can create and save your own presets which is definitely the way to go for UV/IR work. This is not a particular recommendation for Luminar at this time because I need to use it a lot more before I can say whether or not it would replace something like Photo Ninja. But the fact that Luminar has some local editing capability is promising.

 

Visible Globemallow (Sphaeralcea sp.), as shot, followed by conversion.

I thought this As Shot version looked pretty good until I got it into Luminar and realized that it was a bit oversaturated causing some details to be obscured.

 

 

Details of structure and surface textures are nicely brought out in Luminar. And the color here is more realistic. Ambient skylight is very soft, so there are no hard contrasts. But I suppose a pinch more contrast would be good for this photo.

 

 

Reflected Ultraviolet Globemallow (Sphaeralcea sp.), as shot, followed by conversion.

I haven't been getting an exact UV white balance with the S1R. This is partly because in-camera WB is a bit difficult when using SB-140 flash. That flash bounces off even the highly diffuse-reflecting Spectralon.  I'll keep trying though. But no matter, as long as the usual horrendous reds/magentas are toned down so that you can see what you're trying to shoot, I can live with that.

 

 

 

In this conversion I know that I definitely need to brighten up the flowers, but I couldn't quite get that going in Luminar just yet. (Need mo' practice.) The false colors and details are nice though. And I easily brushed in some luminance denoising onto the background which was needed for that black velvet photographed at ISO-800.

As an experiment on these S1R files I sharpened the files beforre reducing from the original whopping 8368x5584 to the posted 1250x1056. Usually I reduce dimensions then sharpen. Just experimenting.

[colinbm]

These look very good Andrea

[dabateman]

Andrea hold on to that Luminar 4 license with an Iron fist. As they just killed it and changed direction again into the machine learning space,  with AI versions. 

I once recommended Luminar 3, but they killed that too. Version 4 isn't as good as version 3 for color adjustments.  Thus why recently bought photo ninja. 

Doesn't photo ninja support the S1R?  If not you could try going through Adobe DNG converter first.

 

[Andy Perrin]

On theoretical grounds it’s always better to sharpen after reducing because sharpening amplifies the high frequencies in the image (which are the edges and the noise) and reducing takes everything higher than a certain frequency called the Nyquist frequency and turns it into an artifact (moire patterns). So you risk moire patterns if you sharpen beforehand. 
 

For the same reasons, it is a good idea to do noise reducing BEFORE reducing the image size, since then you only get sharper edges, rather than sharper edges and more noise/moires. 

[Andrea B.]

Spot on, Andy.

 

*****

 

Bummer to hear about Luminar.

No, Photo Ninja does not yet support the S1R. I need to contact them if I can ever find the time!

 

[dabateman]

Andrea,

I just tested the DNG way and it works fine in Photo ninja.

 

Go here and download the newest free Adobe DNG converter:

https://helpx.adobe.com/camera-raw/using/adobe-dng-converter.html

 

I then downloaded the S1R samples from here:

https://www.imaging-resource.com/PRODS/panasonic-s1r/panasonic-s1rA7.HTM

 

But you already have them. The .rw2 files don't load in Photo ninja as you know. It looked like a solid block of magenta. But the .dng converted files are perfect. 

I didn't change any of the DNG settings, just used the default. 

 

Side Note: The Fuji GFX 50 mk2 samples I downloaded  from here:

https://www.imaging-resource.com/PRODS/fuji-gfx-50s-ii/fuji-gfx-50s-iiGALLERY.HTM

 

and color charts here:

https://www.imaging-resource.com/PRODS/fuji-gfx-50s-ii/fuji-gfx-50s-iiA7.HTM

 

opened fine in Photo Ninja without any DNG conversion needed. 

 

All the above was tested in photo ninja 1.4.0d

[Andrea B.]

Dave, thanks mucho for testing this out! I will def get that DNG converter and use that.

 

Making a quick WB in Silky Pix and outputting a TIF also works, but is not very efficient.

 

In the past I have make simple TIF conversions in Photo Mechanic for use in Photo Ninja, BUT --given that PMechanic makes a TIF from the internal JPG -- if the internal JPG is not saved at full-size, then you naturally don't get a full-sized TIF from PMechanic. For some reason my S1R does not have full sized internal JPGs. Perhaps that is some setting in the S1R? I need to figure that out. Haven't done so yet because there's been so danged much to do with forum software. Well, I'm rattling on here...

[Andrea B.]

As this is an ongoing review of my use of the S1R for UV/IR work, I'll add some occasional note, in no particular order, about some of the settings I've tried.

 

Diffraction Compensation

Camera Icon > Image Quality 2 > Diffraction Compensation

Supposedly adds some kind of acutance(?) enhancement for narrow aperture settings where diffraction might begin to kick in. This cannot be set when there is a manual focus lens in use. Strange. But it probably doesn't matter because a converter's details or sharpening tools can do the same, I suppose.

---

 

HLG Photo

Camera Icon > Image Quality 1 > HLG Photo

THis is a setting to produce a file with increased dynamic range. But the output is an HSP file which can only be viewed on your television. Again, strange. I saw mentioned somewhere that HLG makes for nice "presentations". Why can't we use it to make nice raw files too?

 

iDynamic Range

Camera >Image Quality 2 > iDynamic Range > {Auto, High, Standard, Low, Off}

The iDynamic Range settings alter exposures slightly to push to the right for highlight detail preservation while simultaneously lightening shadows for shadow detail preservation. As an informal quick comment, I think that the Nikon DSLRs do this slightly better than the S1R. But note that I have not hard-tested that statement. Anyway, iDynamic Range is still a useful setting on the S1R. That lightening of dark areas in-camera can be extremely useful in UV photos to bring out details without destroying the light/dark ratios.

 

Cinelike D Photo Style

Camera Icon > Photo Style > Cinelike D

There is an internal curve used lift shadows and keep highlights under control in order to get a good dynamic range. This setting is there mostly for videos, but useful for photos also.

 

Like709 Photo Style

Camera Icon > Photo Style > Like709

This photo style restrains highlights by compressing the right-hand "knee" of the histogram. The illumination slider in Photo Ninja does the same thing. Again this is primarily a video setting, but I find it very nice for photos too. This setting supports exposure to the right.

 

Test Photos to Make

That Lazy Person Andrea B. will probably make these test photos in Visual Mode for a start because I need to make them outdoors in our current bright, contrasty light on windy day. Wrangling UV in that scenario is beyond my current will power.

• One at all 5 iDynamic Range settings with a basic Standard Photo Style.
  All made for a bright, contrasty scene.
• Cinelike D Photo Style and Like709 Photo Style as compared with basic Standard Photo Style.
• Combo of Cinelike D Photo Style with and without an iDynamic Range setting.

 

Photo Style settings range from -5 to +5. Usually for Standard Photo Style, I boost as follows:  +2 Contrast, +3 Saturation, +4 Sharpness. As that is a bit snapshotty for testing, I think I will test with 0 Contrast and 0 Saturation, but retain the +4 Sharpness. Similarly for Cinelike D and Like709 Photo Styles, I will only boost +4 Sharpness.

 

PHOTOS HERE SOON I HOPE

[dabateman]

@Admin Andrea,

Did you ever take an image in full sunlight with the Nikon D610 with UV Nikkor and Badder Venus u filter.  Then pop the lens off, holding the aperture the same and snap the same target with the Panasonic S1r?

I really want to know if the exposure settings are identical between the standard Sony CFA sensor (D610) and the Panasonic S1r with red/ blue color pallet. 

Thanks

[Andrea B.]

Update:   Jim Christian will be adding the Panasonic S1R to the next release V1.4.1 of Photo Ninja. Hurray!!!

---

 

 

David, I'll make that test and let you know.

 

LATER:  It was interesting making that test because I think there are some variables which might affect the results but which we cannot control.

 

Basically, yes, the same aperture and exposure time worked for both cameras when I used their base ISO of 100. I only made a few casual shots (with the Coastal Optics 60/4.0). For a couple of photos, the D610 seemed to need a 1/3 stop faster.

 

But ISO 100 may not necessarily be the same on each camera. Also the conversion replacement glass might not be the same on each camera. As well as the unremoveable sensor covering. Then there is the issue of white balance. Even though you might set the WB for each camera against the same standard, we don't know what the actual temps are. (WB can affect how we set exposure lengths.)

 

So there's some stuff there which gets in the way of any firm conclusions.

[dabateman]

Thank you Andrea,

Assuming exactly the same conversion glass on sensor,  the S1R being a 1/3 of a stop worst case slower is about what I was thinking.  Based on the dye response you found. At 365nm/370nm, you will be mostly green and red on the Nikon. That is 3/4 of the sensor.  The Panasonic s1r will be mostly red and blue, which is 1/2 the sensor.  So should be roughly 1/4 stop less sensitive than the Nikon.

[Andrea B.]

I am still trying to decide what to do with this S1R conversion.

Photosite information for the S1R.

• Number of pixels: 43.7 MP
• Pixel pitch: 4.27 ㎛
• Pixel area: 18.23 ㎛²
• Pixel density: 5.47 MP/cm²

Photosite info for the Nikon D610.

• Number of pixels: 24.3 MP
• Pixel pitch: 5.95 ㎛
• Pixel area: 35.4 ㎛²
• Pixel density: 2.83 MP/cm²

It used to be said that bigger photosites meant more light and less noise. But I don't think that seems to be the case anymore because the S1R has better high ISO capability than does the D610. Of course, there have also been improvements in sensors since the introduction of the D610.

 

---

 

 

 

Today I made a comparative white balance set for a UV photo of a Globemallow flower made with the S1R. It has two UV false colors: UV-reflective yellow petals with a UV-darker blue center and reproductive parts. The stems, buds and leaves are UV-absorbing.

 

Gear [S1R-conversion + UV-Nikkor + BaaderU UV-Pass Filter + SB-140 UV-flash]

 

In the following conversions I tried to demosaic and white balance only. But these converters all have their little ways of doing things so I'm not sure I caught and turned off all the default sharpening.

 

Here is the photo as I shot it, i.e., the JPG version rendered by the S1R's JPG engine. Look about halfway down the flower stem to see a cluster of 3 buds above a leaf on the left side. The frontmost bud is the area I used for all the subsequent white balances. This is NOT how I usually set white balance. But I wanted to make a relative comparison of conversion results by using one particular area of the photo. 

 

 

This is the raw composite from Raw Digger which is demosaiced only with no white balance made on the file. The S1R records more in the blue channel than the D610.

 

 

Here is Raw Digger's auto white balance which is based on the characteristics of the S1R (its white point, for example). This is a pretty rendition even though it is not quite the same as the "standardized" UV white balances we are used to seeing and using on UVP. The blue-violet and yellow are there, but there is an over all purplish cast.

 

 

Here is a Darktable conversion. This version has an overall greenish cast. And the the petals are yellow-green instead of yellow. I scarcely know what I'm doing in Darktable, so I mighta done something wrong!

 

 

Now here is the Luminar conversion. Luminar handles the lights & darks nicely, but leaves the file with a kind of brownish-magenta color cast. Sometimes with Luminar I get a completely standardized UV false blue/yellow/gray palette. Other times I get color casts. Go figure.

 

 

DxO Photo Lab was a huge disappointment. It cannot white balance a UV file at all. It doesn't even get in the ball park with this S1R example! I actually paid for Photo Lab because I wanted to try out their (very very complex) version of "color points". But it will be no use to me if it will not make a white balance on a reflected UV file. The temperature scale goes down to 2000 so I cannot quite figure out why it won't work, but it doesn't. I tried making a WB in other places besides those buds, but nothing worked.

 

 

My last example is a pretty conversion made with the S1R's native converter, a version of Silky Pix. This version probably comes closest to matching our (so-called) standard. I want to make the observation that Silky Pix can handle the files made with a BaaderU, but I you should  know that I have had some UV WB failures in Silky Pix with other gear.

 

 

As I mentioned above, hitting on a particular area of a photo is not the best way to white balance. It is best to use a slab of PTFE which is stable under most UV, Visible and IR light. PTFE fluorescence worries me a bit, but that and Spectralon are all we've got if we want to standardize our white balances. (Not mandatory, of course.)

 

For the record, neither my Raw Photo Processor version nor Photo Ninja 1.4.0d can read S1R raw files. So those two apps are not in the test. 

[Andrea B.]

Testing the S1R Post-Focus Recording & Stacking

page 168-175 in the Lumix S1R Owner's Manual

 

Short Version:  A burst of photos is recorded at (up to) 49 different auto-focused positions.

(Think of a 7x7 grid over the subject being photographed.)

Various combinations of the 49 focus areas are then stacked to make the final photo.

• One photo may be selected from the burst by touching one of the 49 areas in the frame which you want to be in focus.
• Or, the burst can be merged into one photo with a range of different areas in focus. (Just typical focus stacking.)
  • Range Merging - All:  All 49 focus areas are used in the stack.
  • Range Merging with selected subset of the 49 focus areas.
  • Auto Merging: The camera decides the best focus areas to use.

The burst is shot as either Fine JPGs of size 3504 x 2336 pixels or size 5184 x 3456 pixels.

 

The first thing I noticed was that some focus areas are sometimes not used. I do not know if this is because auto-focus failed in those areas or what. In 7 test scenes, only 2 managed to auto-focus in all 49 of the areas. In the other 5 tests, 1,2 or 3 areas of focus were missed. Maybe the light was not good? Maybe this in-camera stacking doesn't work so well on a small subject? More tests are needed with different subjects to figure out why this happened. 

 

The next thing I learned was that Range Merging with All the 49 focus areas did not work well at all. Here are the 7 examples all made with Range Merging - All. They all show really bad merging artifacts.

Frames were resized.

 

You do have to wonder how this in-camera software got out the door? It is truly awful when set to Range Merging - All.

 

 

 

 

The Auto-Merge stacking did better, but the camera did not seem to want to choose to have the tip of this flower in focus except for the 3rd one, a side view, the best of the lot. But I could almost make that 3rd photo without any stacking needed because a lot of the flower and stem are already in the plane of focus. And the algorithm was rather arbitrary about choosing to have the little glass jar in focus - why in only the last frame but not the three preceding frames?

 

 

 

 

 

 

 

[colinbm]

The programmers missed the mark.

[Andy Perrin]

I would agree with colin. There are some very poor results here. Whatever camera dancingcat is using seems to do much better.

[dabateman]

2 hours ago, Andy Perrin said:

I would agree with colin. There are some very poor results here. Whatever camera dancingcat is using seems to do much better.

That's an Olympus Em1mk2 full spectrum,  if I am remembering correctly. 

The Olympus in camera focus stacking is excellent.  I have even gotten it to work handheld with an Em5mk2 a couple of times.