Panasonic Lumix cameras & Post Focus

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: Pau, rjlittlefield, ChrisR, Chris S.

gardenersassistant
Posts: 133
Joined: Sun May 31, 2009 5:21 am
Location: North Somerset, England

Post by gardenersassistant »

Lou Jost wrote:Nick, I was writing about focus bracketing, not post focusing. I am sure that post focus can do better in this regard.
Because of the faster capture rate of post focus? I don't know what the Olympus capture rate is for focus bracketing, but it is about 4.5 per second on my Panasonics, so around 6 times slower than post focus.
Nick

https://www.flickr.com/photos/gardeners ... ollections

Rework and reposts of my images posted in this forum are always welcome, especially if they come with an explanation of what you did and how you did it.

Lou Jost
Posts: 4725
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Yes Nick, that's right. Deep stacks are especially slow, because the buffer fills up and write speed becomes important, though this could be reduced by choosing to skip writing the RAW files.

gardenersassistant
Posts: 133
Joined: Sun May 31, 2009 5:21 am
Location: North Somerset, England

Post by gardenersassistant »

Lou Jost wrote:Yes Nick, that's right. Deep stacks are especially slow, because the buffer fills up and write speed becomes important, though this could be reduced by choosing to skip writing the RAW files.
Oh yes, I forgot about the buffer filling up. I only did a few focus brackets before turning to post focus, and for the few that I did do I quickly limited the number of shots to no more than 45 or so, which is where the G80 buffer fills up.

I'm very content to use JPEG from video now with the G9. I'm getting results I like with Cinelike D. Mind you, I could use that for focus bracketing too, but with 6K giving me 18mpix JPEGs from the 20mpix sensor (as against 8mpix for 4K), and the G9 JPEGs seeming to be fine to me (especially as I'm being careful about setting white balance from a grey card), I'm not sure I would see much benefit from focus bracketing.

And having just one file per "take" rather than loads of individual files is much more convenient.
Nick

https://www.flickr.com/photos/gardeners ... ollections

Rework and reposts of my images posted in this forum are always welcome, especially if they come with an explanation of what you did and how you did it.

Lou Jost
Posts: 4725
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

The G80 is definitely more advanced than the Panasonic cameras I had used, which only did 4k.

gardenersassistant
Posts: 133
Joined: Sun May 31, 2009 5:21 am
Location: North Somerset, England

Post by gardenersassistant »

Lou Jost wrote:The G80 is definitely more advanced than the Panasonic cameras I had used, which only did 4k.
I was a bit unclear in the above. It needs some context (see below). The G80 only does 4K. The G9 does 6K (and Cinelike D).

I started out doing a few 4K post focus stacks with my FZ330 bridge camera, then did more, on and off (mainly off) with a G80. Most of the time I was using aperture bracketing with the G80 so I could choose the balance between DOF coverage of the subject and background rendition at leisure as part of post processing rather than trying to work it out in the field. I didn't use stacking much because of the problems I had with halos around subjects and noise and posterisation in backgrounds. In the months before I got the G9 I did a bit more stacking as I slowly came to better understand how to get it to work.

At around the same time I got an A7ii to see if I could produce botanical images I liked better with that, but that didn't work out so I decided to concentrate on m43 and got a G9. Since then I have been doing 6K stacking for botanical subjects almost all the time, now preferring it to aperture bracketing.
Nick

https://www.flickr.com/photos/gardeners ... ollections

Rework and reposts of my images posted in this forum are always welcome, especially if they come with an explanation of what you did and how you did it.

palea
Posts: 34
Joined: Mon Dec 17, 2018 8:09 pm

Post by palea »

gardenersassistant wrote:"Method A computes the weight for each pixel based on its contrast and then forms the weighted average of all pixels from all source images. This method works better for short stacks and preserves contrast and color."
Good point; I'd forgotten about method A as I didn't use it much when I was working in Helicon. It'd be interesting to see what fraction of pixels are a fairly even combination of images and therefore receive effective noise averaging. For subject pixels I can't really see cases where more than two images would be selected and a majority of such pixels would presumably be closer to one image, which is close to noise neutral. It occurs to me the depth map could be used to locate background pixels and these could potentially be averaged from many frames in the stack. From what I recall of method A it didn't seem to be that smart, though it has been a while.
Lou Jost wrote:With Olympus one selects an easily-accessible menu item and then just presses the shutter.
That's identical to the Panasonic and Nikon implementations. The introduction of post focus and focus bracketing in u43 is typical technological leapfrogging between manufacturers and body generations. Like any variable data stream one can cherry pick the sequence in order to make oversimplified statements. I'm unsure that's the most helpful approach, however. If both features are regularly needed in camera the current choices I'm aware of are a G80, G9, GH5, or use of multiple bodies. (It's possible my choice not to pursue any of those options implies something about marginal utility.)

As @gardenersassistant's pointed out, Helicon's pretty good about handling object translation from wind. If I go back to it, it'll be for this reason. More generally, it's consistently been my experience stacks with lower total motion are more successful. If the process is quick enough to be performed during the low motion period when a plant (or plant borne) subject is reversing direction the advantage can be greater than what a simple comparison of fps might suggest.

I don't see value in reiterating discussion of gaps up thread. We can take anomalous cases up again if @descall comes back with substantive data.
gardenersassistant wrote:The G80 only does 4K. The G9 does 6K (and Cinelike D).
(Minor point of clarity: Cinelike D video is available starting with the G7 and GH4. It's not a post focus option until the G9 and GH5.)

TheDocAUS
Posts: 97
Joined: Thu Jul 19, 2018 10:44 pm
Location: Sydney

Post by TheDocAUS »

Part of my Post Focus adventure means I have finally done some indoor work. I used back-lighting on leaves to see their structure.

I used a cheap LED light box (underside of leaf only) and photographed the leaf from above. I also used exposure compensation to help keep the background white and achieve better details from the leaf (from memory about 2-3 stops over). Here is a Hydrangea leaf. I have not attempted to remove natural defects in the leaf.

In post I made the background pure white, adjusted levels to bring out detail and added some sharpening.

Image

TheDocAUS
Posts: 97
Joined: Thu Jul 19, 2018 10:44 pm
Location: Sydney

Post by TheDocAUS »

I have now established the equipment and workflow for Post Focus in the field. My objective is to get good results efficiently, not the best of the best. I have been photographing a lot of barks and leaves around 2X to 3X. The observations work with those subjects.

This is what I do.

MY EQUIPMENT
• Panasonic G9 (MFT camera), only Panasonic offers Post Focus;
• 30mm and 60mm macro lens (mostly the 30mm, but 60mm is great as well);
• Raynox DCR150 and DCR250. I use the DCR150 occasionally and the DCR250 rarely (as Post Focus struggles with the DCR250 on);
• Wired remote trigger (using the Panasonic app on your mobile phone blacks out the rear screen on the camera). I went back to the wired remote to remove some movement in challenging stacks;
• Sunwayfoto GH-PRO Geared Head (2017 version) on a normal tripod or the ballhead on the flexible K&F Concept TM2534T tripod;
• Sokani X21 Compact LED Video Light mounted on the flash shoe for light. You cannot use flash with Post Focus, only continuous light. The added light definitely improves the final images colour verse natural light.

TECHNIQUE
• Set up tripod with the remote release and turn on LED light;
• Compose the image;
• Take 2 or 3 videos using the Post Focus feature. That way you can choose the best stack later. With close in images (2cm or so from the lens) you can have fine detail like a spider web that moves, even with little or no wind. Often these only become visible when stacking; and
• Stand still so as not to move the tripod, even slightly;
• In strong sunlight I use a transparent white reflector to diffuse the sun, unless I want a high contrast image. Diffused light gives better stacks in strong light.

POST PROCESSING
• Download 6K videos to your PC;
• Extract images from the 6K video using Helicon Focus (by selecting Open Video) The extracted jpeg frames can be found in the Helicon cache;
• Create a test stack using Method C. Method A also works if a stack looks too sharp. Helicon Focus is much quicker than Zerene, so test stacks are done with Helicon;
• 6K stacks give you a TIFF file around 50 megs with 8 bit colour and double that for 16 bit colour. I may use a 16 bit file to process the image in PhotoLemur 3 (see below), but then save in 8 bit once that is done;
• I also have Zerene Stacker for more challenging stacks. Zerene cannot extract frames from a 6K video, so I load the extracted frames from the Helicon cache (you must keep Helicon open until after Zerene finishes its stack, as Helicon deletes the cache when closed);
• Choose the best stack, it can vary between different barks;
• Save the stack in your preferred format. I use TIFF;
• Process the image using PhotoLemur 3, which uses AI and is surprising good. I may make occasional levels adjustments in PS;
• Sharpen the stack. This is where my approach varies greatly. Almost every stack has be sharpened differently, especially with barks. Some stacks are so sharp I reduce sharpness when resizing. Other times I use Franzis Sharpen (great for backlit leaves) and others Topaz Sharpen AI (which I have as a free upgrade). It was this area which took me the longest to work out and I do not have any general guidance, I just need to experiment for each bark type. Apply sharpening only when looking at the image at 100%.

LIMITATIONS
There are limitations to using Post Focus. These are some I have identified so far:
• Post Focus can struggle above 3X, it can work but technique must be first class;
• The AF on the camera can limit what you do. You must be careful to compose so all the important parts of the final image are in the AF areas (so avoid the outer 10% of the visible area on the rear screen);
• Sometimes the AF points are not close enough together to get detail when you are close to your subject and there is, relatively, a deep depth of field;
• If an image has very fine detail (feathers, butterfly wings, etc), the camera can give a blurry image, as the AF system cannot pick up all the fine detail and capture the needed images. There is no manual override for Post Focus (but you could use the Focus Stacking function instead); and
• The AF system appears to struggle with smooth textures. I thought it was the lack of colour contrast, but even high contrast colourful images have the issue, if the surface is smooth. This observation is made concerning barks.

The Atlas of Living Australia (a CSIRO Project), where many images have been uploaded, may write an article on my contribution. The CSIRO was fascinated by the detail of the images and the approach I took. I will post a link whenever it publishes on their blog: https://www.ala.org.au/category/blogs-news/

Regards and happy shooting.

davholla
Posts: 181
Joined: Tue Apr 26, 2016 4:05 am

Post by davholla »

Would this work with Zerene stacker? I have Zerene stacker but not Helicon.

TheDocAUS
Posts: 97
Joined: Thu Jul 19, 2018 10:44 pm
Location: Sydney

Post by TheDocAUS »

Zerene Stacker cannot extract the frames from a 4K or 6K video. You will need one of the utilities mentioned above like FFmpeg to do that.

In fact Helcion uses FFmpeg to extract the frames. The developers just made it seamless, perhaps you can ask Zerene to do the same? I also have Zerene and that would be useful.

Once the frames are extracted as jpegs Zerene Stacker is fine.

TheDocAUS
Posts: 97
Joined: Thu Jul 19, 2018 10:44 pm
Location: Sydney

Post by TheDocAUS »

I have been comparing older manual stacks I have done with the Post Focus images on the G9. The Post Focus images on average are better and sharper, plus much quicker to make and not as prone to wind movement.

Some of the older stacks were done while I was learning, but even the better ones are not as sharp. It seems focus peaking, which I used to help with focus, is not as precise as I thought. I needed more images than Focus Peaking suggested.

It seems this thread is getting lots and lots of views, over 34,000 now. Lots of lookers but few posters. Beginners should feel free to post and ask for help.

TheDocAUS
Posts: 97
Joined: Thu Jul 19, 2018 10:44 pm
Location: Sydney

Post by TheDocAUS »

Received a small mention over at the Atlas of Living Australia for some images I submitted (1,200 and counting). Most of those barks and seed pods referred too used Post Focus with the G9: https://www.ala.org.au/blogs-news/spotl ... tributors/

The endangered mammals used more traditional techniques.

palea
Posts: 34
Joined: Mon Dec 17, 2018 8:09 pm

Post by palea »

Hi, I have some follow on questions about how post focus compares to focus bracketing. These are motivated by curiosity over tradeoffs which might favor one method or another in certain circumstances.
  • It's mentioned earlier in this thread that brackets are displayed as they progress. Does this mean brackets proceed as middle speed bursts or do they get treated as high speed AF-C/AFF bursts? For the G9 and GH5 this would presumably be a difference between 7 fps and 9 fps. The G80/85 and G90/95 are specified at 6 fps either way. These are all substantially slower than the 30 fps of post focus but it's useful to know by how much. However, this isn't something I can currently measure as the body I have doesn't support focus bracketing.
  • It's also mentioned earlier that deep bursts slow down. Panasonic specifies burst sizes as 1) 60 jpeg+raw frames (G9 and GH5), 45 jpeg+raw (G80), or 30 jpeg+raw (G90) or 2) 600 (G90, G9, GH5) or 300 jpegs (G80) with a note in the manuals that the burst may slow down halfway. I'm interpreting this as a reduction from 6-9 fps after 15-30 jpeg+raw or 150-300 jpeg but haven't been able to locate any measurements of how much the slowdown is.
  • The manuals also indicate picture size as a burst speed factor. Has anyone looked at focus bracketing speed with medium size (10MP on 20MP sensors, 8MP on 16MP) rather than large/full size? Medium seems potentially interesting as it's comparable the 8.3MP of 4k frames but provides more bits per frame (less lossy compression and thus potentially higher image quality) while potentially avoiding mid-bracket slowdowns. Since 4k crop is avoided the effective pixel size is a bit larger, suggesting diffraction would be slightly less of an issue for a given effective aperture and slightly less noise would be present for focus stacking to amplify.
For example, suppose one has a 300 frame stack set up but not enough light to meet the 1/30 shutter of post focus. A 1/8 shutter relaxes the light requirement by two stops but still runs the bracket nearly as fast as possible since the burst fps remains close to the shutter speed. Additionally, in a jpeg workflow, frame resolution could be moved from 4k "jpeg" to 16-20MP jpeg and the bracket would still probably complete near maximum burst rate. Alternatively, if bracket file size and stacking speed were the priority, medium picture size could be used.

I've also a corollary question:
  • How does the step size of post focus compare to the step size of focus bracketing? I'm guessing post focus uses step size 1 and that, in principle, focus bracketing wouldn't offer any closer spacing. I still haven't replicated the depth skipping reported with the Olympus 60mm earlier in this thread but, when using microscope objectives, I am finding what may be limitations in what depths autofocus points can pick out for post focus to use. The result is occasional double size steps. Since focus brackets don't try to be smart by skipping empty depths like post focus, they might avoid this.
With certain complex macrographic subjects the apparent double steps result in depth ambiguities due to blur bands. Spending more time collecting a bracket could therefore be faster overall due to reduced retouching time after stacking. However, the actual problem might also be linear autofocus motors occasionally taking larger step sizes like we've seen with stepper motors. If so, focus bracketing presumably wouldn't be better than post focus unless it allows smaller step sizes.

Lou Jost
Posts: 4725
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

I'm guessing post focus uses step size 1
Is that true? I don't think there is a constant step size in Post Focusing.

In focus bracketing, I don't see any need for smaller steps. When using focus bracketing with a microscope objective, he limiting factor is insufficient focus throw on the tube lens. Using a macro lens helps but this pushes down most objectives to unacceptable levels, causing vignetting.

TheDocAUS
Posts: 97
Joined: Thu Jul 19, 2018 10:44 pm
Location: Sydney

Post by TheDocAUS »

Post Focus is a fully automated process using a specialised 4K or 6K video mode on the camera. There are no precise details on what the camera does, that I have found. I have made 1,000s of Post Focus videos now. I can say this by way of observation (I only use 6K) after you press the shutter:

1. The camera works out the focus points in the composition (you see the green AF spots move around the screen). If there are depth changes not picked up by the AF points the camera can miss data giving you a less than perfect stack (this mostly happens around the border of the composition which has no AF points and sometimes in the image).

2. it then takes a 4K or 6K video (as selected by the user) at each AF point worked out in point 1.

3. Most videos take around 2 seconds from when you press the shutter button. Most videos I do are between 30 and 120 frames. All videos are a multiple of 15 frames starting at 30 frames. For example 30, 45, 60 frames, etc Only once did I see a video which was not a multiple of 15 (probably an error).

4. The frames are not taken from front to back in the 6K video. They can mixed up, for reasons I do not know.

5. I have taken a small number of videos over 300 frames, these take longer in camera up to 5 seconds.

6. From what I have seen the Post Focus mode is using 6K/60p to make the stack in Post Focus.

7. I have not taken any stacks since I recently upgraded to version 2 of the firmware.

Post Reply Previous topicNext topic