The Panasonic 100-300 can be extremely sharp

fischem

Member
Messages
11
Reaction score
9
Location
Toronto, CA
Hi

This is my first post here.

Wanted to share this picture I took last night with the Panasonic 100-300:

Panasonic G6, with 100-300 lens, stack of 600 video frames
Panasonic G6, with 100-300 lens, stack of 600 video frames

View: original size

I think this picture speaks to the potential sharpness of this lens.

This is from a 2 min video during which the moon moved almost through the entire frame.

Settings:

Panasonic G6 in manual video setting on cheap tripod
  • f7.1, 1/400 sec, ISO 400
  • Photo Style Natural -2, -2, 0, -5
  • Focus: AFS, AF Tracking
  • lens at 300mm, OIS on
  • AVCHD 1080 60p
  • ETC mode
Processing:

AVCHD Video decompressed and cropped with Virtualdub, saved onto a fast SSD (The 7000 frame AVI file is about 14GB)

I used Emil Kraaikamp's brilliant Autostakkert to analyse the video and stack the sharpest 8% of frames. This resulted in a extremely high SNR 16bit TIFF file. Sharpened with Photoshop smart sharpen.

This type of photography is common among Astro photographers and is also called lucky imaging. It delivers results that are very close to the optimal performance of an optical instrument.

Other observations

I've had mixed results with this lens before, but as others have said it's not due to the optical quality, but mostly because of shake or shutter shock. I now use the electronic shutter almost exclusively and go for fast shutter speeds. Some of the shots of wild life in Costa Rica have turned out quite good - I love the reach of this lens!

Martin Fischer

Toronto

PS: I posted this by accident on the Panasonic forum earlier
 
Looks good. Definitely stacking the video frames is the way to go.
 
I use the 100-300 a lot and find it to be very sharp. My usual 'sharp' f stop is 7.2 and it always delivers very good detail. By the way, I use it on Panasonic and Olympus bodies...no difference in the behavior of the lens.
 
Thanks mpgxsvcd!

Here's the logic behind the stacking:

In a single photo, small features with little contrast don't look any different from noise and are therefore drowned out. By stacking a large number of frames the noise gets reduced and only the features captured by the optics and sensor remain (apparently the Signal to Noise ratio increases with the square root of the number of frames, see deepskystacker.free.fr). When capturing the video for this picture, I had set the Noise Reduction to the smallest value, so that those small features would not get deleted by the noise reduction algorithm.

Have a look at the comparisons below. Some features such as the Alpine Valley are not even visible in the single frame.

left: stack of best 8% of 7100 frames. Right: single image
left: stack of best 8% of 7100 frames. Right: single image

left: stack of best 8% of 7100 frames. Right: single image
left: stack of best 8% of 7100 frames. Right: single image

Also, Autostakkert ranks all the frames according to sharpness, which allows to use only the very sharpest frames (hence the term "Lucky Imaging").

So, at the end of the day, this method allows the user the get the most out of their optics, especially when shooting with long focal length through turbulent air.

For me, it was just proof that I could get very sharp pictures with the Panasonic 100-300 lens if I do everything right on the operational side.

Martin

PS: Be aware that the stacking software delivers a picture that looks very blurry. But modern sharpening tools as Photoshop's smart sharpen filter can apply deconvolution. To do that in Photoshop check the "more accurate" box. Another software that is great for sharpening (and stacking) is the freeware Registax.
 
That does look pretty sharp. Is what you posted the limit of pixel resolution that you can get from this method (starting with video), or did you downsize? (Not to nitpick, but it does look a tad over sharpened. Maybe that is just something you have to live with with this technique.)

It is counter-intuitive that you could get more resolution from video than from using the full frame resolution. Why not stack still images, at a higher resolution to start? Or is this just a practical matter, easier to get large numbers of images from video.

Could you say something about whether the stacking actually can recover detail at a higher resolution than the single-frame resolution would suggest? I'm kind of thinking you can, but I haven't though enough about it to be certain, or clear on how it works and what its limits are.

I LOVE the astro community and what they do for/with photography. Lately started using a couple of apps from them, and I find them brilliantly conceived and executed, with just the right bells and whistles that I need, without feature bloat that you find in many "more professional" apps. Bespeaks knowing the technology and the math, and a good design aesthetic, to my taste (having been a software designer of sorts, coming from a technical background).
 
Very interesting example and explanation. I have to try stacking one of these days.

On a mundane level I use this lens a lot, primarily for HH snapping birds - as much for identification as any hopes of "art". Two things have increased my initially poor keeper ratio. One is setting the anti-shock to, as far as I recall, 1/8th. A value that seems idiotic but which I adopted from someone else's recommendation. The other improvement comes from using the continuous shutter release - high seems best but low works too.

Roy
 
I wouldn't say 'extremely' sharp, but it is definitely SHARP. And I think it's definitely usable by all kinds of photographers and enthusiasts. As far as I'm concerned, the 100-300 should be in every 43 shooter's arsenal of lenses.
 
boxerman wrote:

That does look pretty sharp. Is what you posted the limit of pixel resolution that you can get from this method (starting with video), or did you downsize? (Not to nitpick, but it does look a tad over sharpened. Maybe that is just something you have to live with with this technique.)
I agree about the over sharpened look. I still work on my sharpening technique - ideally sharpening would bring out the small features (such as ejecta from the craters) without exaggerating the high contrast features such as crater rims. I tried to find a middle ground. If anyone know how to do this, I would love to learn!
It is counter-intuitive that you could get more resolution from video than from using the full frame resolution. Why not stack still images, at a higher resolution to start? Or is this just a practical matter, easier to get large numbers of images from video.
On most Panasonic G series cameras (GF, G or GH), there is the option to shoot video in Extended Tele Conversion mode, which only reads data from the center 1920x1080pixels of the imaging sensor. This 2.5x crop factor turns the Panasonic 100-300 into a 500-1500mm zoom lens in terms of 35mm film.

When shooting in this ETC mode, the moon covers exactly the same amount of pixels as it does when shooting 16mp still images. The advantage of video is that you can get 60 frames per second, or 3600 frames per minute. It also just so happens that the moon fits nicely into the video frame. The advantage of still images that you can shoot in raw, which means that the raw data from the sensor is available. For deep sky imaging, such as nebulas and galaxies raw files are preferable. With the new Blackmagic video cameras, one could have the best of both worlds: Thousands of raw frames in a matter of minutes....
Could you say something about whether the stacking actually can recover detail at a higher resolution than the single-frame resolution would suggest? I'm kind of thinking you can, but I haven't though enough about it to be certain, or clear on how it works and what its limits are.
Yes, that is called super-resolution. The Autostakkert software I used has the option to do a 1.5x or 3x enlargement method called drizzling. The author of that software has an interesting article on that: Enhance. In our case though, there might not be gained too much from this, since there is an anti-aliasing filter in front of the sensor of all Panasonic's cameras, which slightly blurs the image. (Not sure if the Nikon 800e, which does not have such a filter, offers a similar crop mode to the Panasonic cameras. Anybody out there knows about this?)
I LOVE the astro community and what they do for/with photography. Lately started using a couple of apps from them, and I find them brilliantly conceived and executed, with just the right bells and whistles that I need, without feature bloat that you find in many "more professional" apps. Bespeaks knowing the technology and the math, and a good design aesthetic, to my taste (having been a software designer of sorts, coming from a technical background).

--
The BoxerMan
 
Thanks for the explanations. Sorry, I can't help with sharpening techniques. Not my strength.

My camera, Olympus E-M5, also has a similar crop-to-magnify feature they call digital teleconversion. (I'm not sure it works in the same way on video, as for stills.) But, for stills, the camera also interpolates so you get apparently the same number of pixels in the image as the non-cropped images, not just same number of pixels on target.

Drizzling seems a bit magic. No time now, but someday I'll figure out the details. Thanks for the reference.
 
Good result.
 

Keyboard shortcuts

Back
Top