In a previous post I wrote about a focus stacking effort I made with images of a model aircraft at a show. I had been meaning to have another go at this and do so in a more controlled environment. I then ended up buying myself a macro lens for use in my negative scanning efforts and immediately started playing with it to shoot things close up – it’s a macro lens for goodness sake!
As an f/2.8 lens, when shooting macro shots, the depth of field is really shallow. This got me thinking about trying another focus stack. A small Leatherman seemed as good a target as anything. I set up with manual focus, put the camera on a tripod, went to manual exposure and then shot a sequence with small changes to the position of focus for each shot. Then it was off to Photoshop.
Photoshop did a pretty good job really. The distortion of the areas out of focus means that the area that the subject covers can vary quite dramatically as the focus shifts backwards and forwards. The algorithm did well getting things masked and blended. The only bit it struggled with was at the very top where the knurled edge seemed to confuse it a bit. The top shot is the finished effort while two others are included to show how much things are out of focus in the individual shots.
I have previously written about some time-lapse software that I use based on a recommendation from my friend Jo Hunter. I used this software for a while before the creator, Gunther Wegner, updated it to a new version and deactivated the version I had. I was a bit miffed by this but I was able to continue using the export settings from his application within Lightroom but with me having to manually manage the file cropping. All transition and smoothing capability was lost but I could still make a basic time-lapse.
Sadly, the latest versions of Lightroom have done away with that as well and now I couldn’t even render the video. Therefore, I decided to take a look at the latest version of his software. It is now on version 3 so I have skipped a version en route to this place. I had shot a few sequences recently and wanted to be able to manage them properly so decided to come back and have another go with his application.
There have been some improvements in tidying up the software so the workflow is a bit better. There is still a certain amount of effort as you switch back and forth with Lightroom. You have to put all of the images into a single folder while working on them. This means a modification to my storage strategy but it isn’t difficult to manage in Lightroom and, when you are done, you can revert the images to their original locations and still render the video output. It might be nicer if he gave you a more flexible approach to selecting files but this is not a hardship.
It has changed a little from what I am used to since now it starts out analyzing the files before you have created keyframes. Once this is done, you save the xmp files out and reload them in Lightroom. Define your keyframes and make any edits to them that you want and save xmp again. Back to LRTimelpase to load the changes and now it does its smoothing very quickly. Save xmp again and back to Lightroom and now there is an Export setting along with all of the other export options. No going to the Slideshow module any more.
The results are pretty good. The new export functionality actually generates a sequence of still images which it stores. You then head back into LRTimelapse where you have a series of options for rendering a video sequence from those stills. You can have it automatically delete the stills when it is done or keep them and render again using different output options. This is pretty flexible although it means you have to pay attention to what you have used so you don’t end up with a ton of stored intermediate images. The video output looks good. I am not using the product commercially so do not have the full commercial license. That will allow output in 4K formats. I only wanted 1080p HD format for my purposes. It is a little interesting that 4K is automatically assumed to be commercial use. With people able to record 4K on GoPros these days and 4K TVs showing up, I suspect this is going to be a more mainstream format before too long. Maybe it will be in the personal use license of future versions of LRTimelapse. We shall see.
Was it worth the upgrade? Yes. Not least because I wanted to have the capability back that I had lost when Lightroom changed its output but, even so, I have found the new version to be quite a bit more friendly to work with. I think it will encourage me to work on a few more projects.
I have been making some shots with multiple exposures to overlay. This is something I have posted about before and the shots here are similar to those from before. However, this post is less about the shots and more about the post processing I used. Previously I opened up all of the shots as layers in a single file and then auto-aligned them. Once done, I then used the Auto Blend functionality to show each shot o the aircraft in place.
This was a lot quicker than my previous approach and was something I picked up from posts on photographing star trails. However, recently, I have not been as happy with the results as I should have been. Some of the planes, particularly those near to the edges, had some odd artifacts appearing. Also, if there were any overlaps, the blending masks could give some weird effects. Therefore, I have taken a different approach for a while. This is slower, I admit, but I think it gives a better result.
Once the alignment of the images is done, I hide them all except the bottom layer by Alt clicking on the eye beside the last layer. Then I add the next layer up back in but mask it out completely. A white brush on the mask then allows me to paint back in the new aircraft positions. This is a bit laborious but it does allow you to decide exactly what you want in and what you don’t. if one file is not helpful to the composition, you can easily ignore it.
If the layers are not all exactly aligned from shooting on a tripod, you will also get gaps at the edges on different layers. You can also fill these in by brushing in the layers that provide the right coverage and get a complete image. Once you are happy, flatten the whole thing and you are done.
Ever since Adobe got their act together with the Photomerge function in Photoshop, it has been my default for creating panoramas. Previous versions were a little unreliable but they cracked it a few versions back and I have not changed my approach since. However, a recent bit of YouTube exploring has changed that again. Russell Brown had a series of videos on making panoramas from aerial shots and, while I was watching them because they were aerials, the stuff he came up with on stitching panos was actually more useful to me.
Instead of using Photomerge, he uses the same functionality of Photoshop but in individual steps. The technique involves opening all of the shots as layers in a single file (something you can do straight from Lightroom which maintains the re-import linkage that I like about Photomerge). Then you select all layers and use Edit>Auto Align Layers. This gives you the same options as Photomerge. Apparently, according to Russell, if you have a series of shots where you have rotated position but are looking horizontally, Spherical works best. If you are looking up or down, use Circular.
The result can be quite distorted if you use Circular but it will all get better soon. If you aren’t happy, you can undo the step which is a lot easier than starting from scratch which is what you would have to do in Photomerge. Once the alignment is done, Edit>Auto Blend Layers while take care of the rest of the stitching and blend everything together nicely. It defaults to a Panorama blending option.
With the blend done, flatten the layers and open Filters>Adaptive Wide Angle. This will default to a panorama setting and, if you have the distorted output from the Circular settings, now you will suddenly see everything come back to what it should be. You can tweak this filter to get verticals and horizontals aligned as you wish and then you end up with a good pano output. Some cropping and filling of blank areas with Content-Aware fill and the job is done. I shall be taking this approach for all my panos from now on. If you don’t use this approach already, you might want to give it a try.
Also, you can go to the original source on this and check out Russell’s videos on YouTube.