Hi all,
Today's wet weather seems to be bringing some new creatures out to play. There were several of these in a corner of my garden in amongst the leaf litter.
I'm not 100% sure that this is a "Black Snake Millipede", Tachypodoiulus niger. It did look very black when on the ground, but seems a lot lighter up close. My field guide says that it is the commonest British millipede.
I was surprised to see how primitive the eye looks. I suppose they don't need much vision when rummaging around in the undergrowth.
The field of view is a little under 2mm in this image.
I re-shot these images three times today, as I couldn't get the mouth parts sharp in the stack. I gave up after three tries and settled for the best I could get. I did try to copy the mouth from the individual frames using masks in Photoshop, but the depth of field here is so small that there were several frames with part of the mouth in focus, so it was difficult to get much from them for the final image.
D200 with Schneider Componon-S reversed on my home made focussing rail.
3 sec exposure @ F.6 using tungsten light
72 frames with 0.02 mm adjustment
Stacked in Helicon Focus
Finished in Photoshop.
Bye for now.
Black Snake Millipede
Moderators: rjlittlefield, ChrisR, Chris S., Pau
-
georgedingwall
- Posts: 207
- Joined: Tue Aug 01, 2006 11:15 am
- Location: Invergordon, Scotland
- Contact:
Black Snake Millipede
Last edited by georgedingwall on Sun Mar 02, 2008 2:11 pm, edited 1 time in total.
-
rjlittlefield
- Site Admin
- Posts: 24147
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
The photographer always knows what's wrong! 
The image looks pretty good to me too.
Where are the mouthparts that are fuzzy? Can you post out a blown-up crop with some sort of pointer?
Also, have you tried the retouching features in Helicon Focus? They're not magic, but they are coordinated with whatever image registration HF had to do. That removes one more thing to go wrong.
HF also has a very nice way of adjusting the brush size using the mouse wheel -- much slicker than Photoshop's making you go back to the brush menu every time.
--Rik
The image looks pretty good to me too.
Where are the mouthparts that are fuzzy? Can you post out a blown-up crop with some sort of pointer?
Also, have you tried the retouching features in Helicon Focus? They're not magic, but they are coordinated with whatever image registration HF had to do. That removes one more thing to go wrong.
HF also has a very nice way of adjusting the brush size using the mouse wheel -- much slicker than Photoshop's making you go back to the brush menu every time.
--Rik
-
georgedingwall
- Posts: 207
- Joined: Tue Aug 01, 2006 11:15 am
- Location: Invergordon, Scotland
- Contact:
Hi Rik,
The area on the tip of the upper beak like structure just wont come sharp. It may be due to the lack of contrast on its surface. Even when you look at the individual frames, it's quite hard to see the area moving in and out of focus.
I briefly tried the "Clone Source" option to clone in areas from the original frames, and this does seem to have possibilities.
Thanks for the thoughts Rik, bye for now.
Here's the two areas I wasn't able to get to be as sharp as I would have liked. At first I thought that the item at the lower jaw was a pair of Palp like structures, similar to what you find on a beetle. I looked at it under the microscope, and it is actually one of the first pair of legs. So the fuzzieness may be the result of the edge artifacts you get in HF when an edge passes in front of something deeper in the field. I might try one final stack to see if I can remove the leg form that part of the image.Where are the mouthparts that are fuzzy? Can you post out a blown-up crop with some sort of pointer?
The area on the tip of the upper beak like structure just wont come sharp. It may be due to the lack of contrast on its surface. Even when you look at the individual frames, it's quite hard to see the area moving in and out of focus.
I've only briefly tried the clone tool, but found it a bit awkward to use. I didn't like having to keep pressing the "Clone New Area" button when you want to change the position of the clone source.Also, have you tried the retouching features in Helicon Focus? They're not magic, but they are coordinated with whatever image registration HF had to do. That removes one more thing to go wrong.
I briefly tried the "Clone Source" option to clone in areas from the original frames, and this does seem to have possibilities.
I had noticed that, and it is a good feature. In photoshop I never use the brush menu to change the brush size. I always use the Square Bracket keys to vary the size of the brush. The left one reduces it and the right one increases it. I'm not sure that I would want the mouse wheel to change the brush size in Photoshop, as I use it at present to zoom in and out and to scroll around the image.HF also has a very nice way of adjusting the brush size using the mouse wheel -- much slicker than Photoshop's making you go back to the brush menu every time.
--Rik
Thanks for the thoughts Rik, bye for now.
Last edited by georgedingwall on Sun Mar 02, 2008 2:11 pm, edited 1 time in total.
-
rjlittlefield
- Site Admin
- Posts: 24147
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
George,
I think you're right that the problem is lack of contrast in those surfaces. The hairs near the pointers look just as sharp as everywhere else in the image. So do the catchlights about 1 cm on screen below the lower arrow. But I agree that most of the surface looks like a featureless beige blob.
A couple of possibilities occur to me.
First, it could be that those surfaces really are featureless beige blobs, in which case there's not much you can do except play around with the lighting to try to highlight some more texture, assuming there is some.
Second, it could be that even the surface texture in those areas is too small to resolve with good contrast. It seems clear from the blowup that you're working at the resolution limit of that lens.
Some improvement in usable resolution can be made by aggressive sharpening, for example using Photoshop's unsharp mask at 100% with 1 pixel radius. What you're trying to do is level up the end of the system's MTF curve by using the unsharp mask to compensate for falloff in the lens's MTF at high spatial frequencies (fine detail). You'll have to experiment with the radius and percentage to see what gives the best compensation. Down here at the limits of resolution, you might need what would otherwise be absurdly large percentages -- 150% or 200% may be completely reasonable.
I've never used "clone new area" in HF. My mindset is that cloning another area is a last-ditch effort to paint in features that look plausible even though they weren't caught by the camera. That bothers the scientist in me -- if the camera didn't catch them, why do I think I'm painting them correctly? When my artistic side requires it anyway, I prefer to work in Photoshop because "step backward" and layers/masks gives me better control.
--Rik
Ah! You're pointing at different things than I expected. (Have I ever mentioned my fondness for "point and grunt" as a communication technique?)Here's the two areas I wasn't able to get to be as sharp as I would have liked.
I think you're right that the problem is lack of contrast in those surfaces. The hairs near the pointers look just as sharp as everywhere else in the image. So do the catchlights about 1 cm on screen below the lower arrow. But I agree that most of the surface looks like a featureless beige blob.
A couple of possibilities occur to me.
First, it could be that those surfaces really are featureless beige blobs, in which case there's not much you can do except play around with the lighting to try to highlight some more texture, assuming there is some.
Second, it could be that even the surface texture in those areas is too small to resolve with good contrast. It seems clear from the blowup that you're working at the resolution limit of that lens.
Some improvement in usable resolution can be made by aggressive sharpening, for example using Photoshop's unsharp mask at 100% with 1 pixel radius. What you're trying to do is level up the end of the system's MTF curve by using the unsharp mask to compensate for falloff in the lens's MTF at high spatial frequencies (fine detail). You'll have to experiment with the radius and percentage to see what gives the best compensation. Down here at the limits of resolution, you might need what would otherwise be absurdly large percentages -- 150% or 200% may be completely reasonable.
"Clone Source" is the tool I had in mind.I've only briefly tried the clone tool, but found it a bit awkward to use. I didn't like having to keep pressing the "Clone New Area" button when you want to change the position of the clone source.
I briefly tried the "Clone Source" option to clone in areas from the original frames, and this does seem to have possibilities.
I've never used "clone new area" in HF. My mindset is that cloning another area is a last-ditch effort to paint in features that look plausible even though they weren't caught by the camera. That bothers the scientist in me -- if the camera didn't catch them, why do I think I'm painting them correctly? When my artistic side requires it anyway, I prefer to work in Photoshop because "step backward" and layers/masks gives me better control.
Excellent -- I had missed that keyboard shortcut. Thanks!In photoshop I never use the brush menu to change the brush size. I always use the Square Bracket keys to vary the size of the brush. The left one reduces it and the right one increases it.
--Rik
-
georgedingwall
- Posts: 207
- Joined: Tue Aug 01, 2006 11:15 am
- Location: Invergordon, Scotland
- Contact:
Hi all,
I've reworked the stack of images using Rik's suggestion of utilizing the retouching features of Helicon Focus.
I mainly used the "Clone Source" option which allows you to clone the final image from the original unstacked images.
I think I've managed to tease out some extra detail and edge definition in the new version than I got in the areas I was concerned about in the old one. I think that with a bit more practice, this could be a useful tool to use when HF appears to lose detail in the final image but which is there in the originals.
Bye for now.
I've reworked the stack of images using Rik's suggestion of utilizing the retouching features of Helicon Focus.
I mainly used the "Clone Source" option which allows you to clone the final image from the original unstacked images.
I think I've managed to tease out some extra detail and edge definition in the new version than I got in the areas I was concerned about in the old one. I think that with a bit more practice, this could be a useful tool to use when HF appears to lose detail in the final image but which is there in the originals.
Bye for now.
Last edited by georgedingwall on Sun Mar 02, 2008 2:11 pm, edited 1 time in total.
-
rjlittlefield
- Site Admin
- Posts: 24147
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
George,
You must have been working on this while I was composing my last post. It's either that, or you're a miracle worker who can do lots of work and post out the results in 5 minutes!
This image does look a lot better than the first one (which looked pretty good, not having the actual critter to compare against).
It's interesting to layer your first & third images, register them, then click the upper layer on/off to quickly toggle between the two images for comparison.
Things I notice are that 1) the area where foreground mouthparts come up against background palp (or whatever that is) seems a lot cleaner in this new image, 2) the general shape of the lower jaw seems different -- it's a lot wider in the first image, 3) there's some geometric warping between the two, as if the point of view changed slightly, 4) there are a few clear differences in fine detail, notably some dimples on the front of the head, just above-right of the antenna, that appear in the first image but not this latest, and 5) the color is a lot more saturated in the latest image.
The point about geometric warping is one that I have not mentioned before, though I've seen it frequently in my own work.
There is an issue is that as software works through the stack, it generally registers images incrementally, each image against (only) the previous one. Errors accumulate in this process, and it's common for the end of the stack to be shifted and/or scaled significantly different from what it should be, given the actual geometry of the subject & optics. This appears in the final image as something like a shift in viewpoint, or sometimes (depending on the object shape) as even a nonlinear warping of the shape.
The shifting/warping is hard to spot, unless you layer-compare two images made with different software from the same stack, or sometimes even the same software with different parameters. Then it jumps right out at you. I first noticed this when comparing Helicon Focus with CombineZ5 with my Panorama Tools hack -- the final geometries were all different!
I'm curious -- were your first and third images generated from the same stack, or different stacks having the same geometry, or different stacks with possibly different geometry?
--Rik
You must have been working on this while I was composing my last post. It's either that, or you're a miracle worker who can do lots of work and post out the results in 5 minutes!
This image does look a lot better than the first one (which looked pretty good, not having the actual critter to compare against).
It's interesting to layer your first & third images, register them, then click the upper layer on/off to quickly toggle between the two images for comparison.
Things I notice are that 1) the area where foreground mouthparts come up against background palp (or whatever that is) seems a lot cleaner in this new image, 2) the general shape of the lower jaw seems different -- it's a lot wider in the first image, 3) there's some geometric warping between the two, as if the point of view changed slightly, 4) there are a few clear differences in fine detail, notably some dimples on the front of the head, just above-right of the antenna, that appear in the first image but not this latest, and 5) the color is a lot more saturated in the latest image.
The point about geometric warping is one that I have not mentioned before, though I've seen it frequently in my own work.
There is an issue is that as software works through the stack, it generally registers images incrementally, each image against (only) the previous one. Errors accumulate in this process, and it's common for the end of the stack to be shifted and/or scaled significantly different from what it should be, given the actual geometry of the subject & optics. This appears in the final image as something like a shift in viewpoint, or sometimes (depending on the object shape) as even a nonlinear warping of the shape.
The shifting/warping is hard to spot, unless you layer-compare two images made with different software from the same stack, or sometimes even the same software with different parameters. Then it jumps right out at you. I first noticed this when comparing Helicon Focus with CombineZ5 with my Panorama Tools hack -- the final geometries were all different!
I'm curious -- were your first and third images generated from the same stack, or different stacks having the same geometry, or different stacks with possibly different geometry?
--Rik
-
georgedingwall
- Posts: 207
- Joined: Tue Aug 01, 2006 11:15 am
- Location: Invergordon, Scotland
- Contact:
Hi Rik,
I had just posted the reworked image when I was notified of an update to the topic. I was a bit surprised, as you don't usually get a notification for your own posts, but when I went to the forum, you contribution was there. The two posts must have shook hands as they passed through a satellite.
The reworked image was made from the same batch of images, but run with slightly different settings. The first used R=4, S=4, and the second used R=8, S=4.
For the source cloning, I used several of the original frames to progressively get a few bits of texture and definition for each where I could. It's possible that I got a little too close to the edge of the jaw when cloning, and pulled in a bit of fuzz which has slightly altered the edge of the jaw. It may just be my lack of skill in using this method that has meant I may have made some unintended edits.
I intend to get more familiar with the "Clone Source" option, as I think it is possible to recover some detail lost in the stacking process.
I did notice when I was selecting the frames for cloning that there was some slight shifting of the frames as I went. I've often noted that, even with the same set of images, you can get some warping from one run to the next.
As for the colour change, that is just me. I can never remember the exact settings I used from one image to the next, so I just go through the motions again and hope that what I end up with is OK. If I absolutely must have the exact same settings, I usually create an action or script to record the settings. I don't do this often, but if I am making a set of prints or images that are to be displayed alongside each other, I would do it then so that I'm sure they are as balanced with each other as possible.
Thanks for the tips today, Rik. I've learnt a few things which should help me produce better images in future.
Bye for now.
I had just posted the reworked image when I was notified of an update to the topic. I was a bit surprised, as you don't usually get a notification for your own posts, but when I went to the forum, you contribution was there. The two posts must have shook hands as they passed through a satellite.
The reworked image was made from the same batch of images, but run with slightly different settings. The first used R=4, S=4, and the second used R=8, S=4.
For the source cloning, I used several of the original frames to progressively get a few bits of texture and definition for each where I could. It's possible that I got a little too close to the edge of the jaw when cloning, and pulled in a bit of fuzz which has slightly altered the edge of the jaw. It may just be my lack of skill in using this method that has meant I may have made some unintended edits.
I intend to get more familiar with the "Clone Source" option, as I think it is possible to recover some detail lost in the stacking process.
I did notice when I was selecting the frames for cloning that there was some slight shifting of the frames as I went. I've often noted that, even with the same set of images, you can get some warping from one run to the next.
As for the colour change, that is just me. I can never remember the exact settings I used from one image to the next, so I just go through the motions again and hope that what I end up with is OK. If I absolutely must have the exact same settings, I usually create an action or script to record the settings. I don't do this often, but if I am making a set of prints or images that are to be displayed alongside each other, I would do it then so that I'm sure they are as balanced with each other as possible.
Thanks for the tips today, Rik. I've learnt a few things which should help me produce better images in future.
Bye for now.
First off, I didn`t even know all that stuff was in helicon Focus 
I see a good difference in the second picture, more texture in the mouth area and more color also. I was going to suggest a change in background color....like a darker gray. A darker color would give the creature more edge contrast and the hairs would show a lot more IMO. Still is a fantastic stack George
I see a good difference in the second picture, more texture in the mouth area and more color also. I was going to suggest a change in background color....like a darker gray. A darker color would give the creature more edge contrast and the hairs would show a lot more IMO. Still is a fantastic stack George Take Nothing but Pictures--Leave Nothing but Footprints.
Doug Breda
Doug Breda