Header image


Planetary Processing My Way

Over the last 4 years I have developed what I think is a sure fire method for planetary processing. Unlike others out there in cyberland I don't believe in having a mulititude of steps, which invariably distort and damage the precious data. For me it is all about keeping it simple. Good data really does not require lots of processing; so capture is really important. If you data is not great, don't bother processing it unless you want to test your skills. Remembering of course that while you are still learning data capture, many times you will have poor quality data. In that event, even these simple rules I work by will aid in your processing.

Some of my simple rules really are fairly common sense. However, let me state that nothing looks worse than an overprocessed image. The idea as I see it, is to produce an image that looks like it was taken from orbit of that planet. I don't think it should look like it is some image that was drawn with pastels. Fine detail is often lost when over processing. So the over riding rule is to make the image look smooth and natural.

REGISTX: ALIGNING, STACKING & WAVELETS (see image and text below)

Open Registax4 and click on the Select tab to select your AVI or SER file. In my case I select on the particular colour I want to process. Next grab hold of the bottom scroll bar and scroll through until you find the sharpest frame possible. Scrolling quickly will show you how the image goes in and out of sharpness because of the seeing. Once you have found the frame you like click on the Alignmentboxsize area and select the 256 box as shown below. Now click on a feature that is close to a high and low contrast area. In the image below you can see I have selected a point in the SEB of Jupiter that has some barges and the darker SEB material in it. I have found that by doing this, the program gets a very good lock on that feature and will select the sharpest frames that look like this.

Now look at the FFT filter and in particular the number at the bottom right. This figure should be between 5-10. If the number displayed is more than this value just change it to be around that figure. Next select the lowest quality in the Quality estimator. I like to use a figure of around 85% and the Method of Gradient. It you have over 1000 frames then these setting will work well. If you are using Toucam AVI's then try some where around 70%. This is because you will have around 500-700 frames and you will need a lot of lesser quality frames to help reduce noise.

Now click Align and let the program sort all the frames out as per what you have set.



Using Limit (see image below)

Once the alignment process is completed then it is time to limit the defective frames from entering the optimization stage. In the image below you will see a graph. The best frames are on the left and as you move to the right the frames slowly degrade in quality. So if you grab the bottom slider again and move it from right to left you can find the minimum acceptable frames you want. Now there is no hard and fast rule here, it is more a thing of judgement. For me I like something like around 1000 frames to enter the optimization stage. I look at the frames and as I get towards 1000 the detail gets better in each frame. This is what I am aiming for, detail and lots of frames. As you get better at processing you will acquire a feel for where to stop.

So when you are happy, click on the Limit button and progress on to Optimization.



Creating a Reference Frame (see both images below)

Now that you have clicked on limit you will get the optimization screen. Go to the "Create reference frame" and click on create. You don't need to change the frames to stack, you should have more than enough frames to work with. However, if you have less frame in you overall stack you can change the amount of frames. This feature uses your top 50 frames to make a frame that the program will use as a guide when it is doing the optimization process. So it is fairly important that you have this frame in your processing.



When the program has stacked the frames for the reference frame it will take you to the wavelets area (see image below). Look at the area marked Processing and click on the box marked as Automatic. There are a variety of controls here that you can manipulate but I just work with the wavelets. I now only use wavelets marked 2 & 3 as shown below. I move both sliders until I get a sharper image with just a small amount of speckle. I want to get the fine data and am not concerned with large blocky looking features. I try not to go above 25 on each slider to bring out the detail. As you can see from the image below there is plenty of detail showing for this stage of the processing. Don't push the data too hard as there are still plenty of steps that involve sharpening yet. Now click Do All and then click on Continue.



Optimization (see image below)

When you click continue in the previous step, the program takes you back to Optimize. Make sure that you have the single run optimizer under Options unchecked. You want the program to completely optimize all the frames in order of quality. Now click Optimize only. Depending on your computers abilities this may take some time. Go off and make a coffee for a few minutes.



Stacking (see image below)

After the frames have been optimized click on the tab marked Stack. You will now be taken to the stacking part of the program. Check the box named Show Stackgraph and the use the sliders in the stack graph to once again limit the bad frames. You can move the verticle one to remove frames that have radical peaks and more importantly slide the horizontal slider from right to left until you start to see frames that are showing plenty of detail. In the image below I have selected around 684 frames. I typically like to use around 980 odd frames as my final selection. When you are happy with the amount of frames you want to use for your final stack simply click the Stack Button.



Wavelets (see image below)

With all your frames now stacked click on the wavelets tab and do your final adjustments. As you can see from the image I have only adjusted wavelets 2 & 3 and just so that the image is smooth but showing plenty of detail. These settings will change from AVI to AVI. No two video's will have the same settings applied to them. Each is a creation of their own. This is more about how the image looks rather than working to some principled ideals.

For those using colour video you need to do an RGB align and you can do a histogram adjustment too. With the RGB align use the estimate function to get your colours correctly aligned. You can do more in the next program that I am going to talk about.

When you are finished, click Do All and then click on the Final Tab.



Saving (see image below)

Once again there is a variety of controls that you can use, but I only want to save my data here. So I click on the Save tab. I like to save in Tiff format and I save it in the same folder as the video data. If you are running monochrome, you need to repeat this process for each colour.



ASTRA IMAGE: SHARPENING AND COMBINING (see images and text below)

Astra image is a valuable tool in my arsenal. It allows you to sharpen your images by using a process called deconvolution and it also allows me to recombine the data to make a colour image.

First thing you need to do is open you RGB saved images from Registax. These will be named red, green and blue. If you are using a colour image from a colour avi, then click on process and click split colours.

With monochrome images convert each frame to greyscale by clicking on process and then clicking on convert to greyscale. Stack each greyscale on top of the associated monochrome image.



Deconvolution (see images below)

Deconvolution is a sharpening process that is much more powerful that wavelets in Registax. However it must be used carefully so as not to destroy the data. Some people use Maximum Entropy (ME) however I like Lucy Richardson (LR). Personally I think LR gives a crisper image and allows fine detail to show. The use of ME typically gives a blocky look to the image where large features are well seen but find whisps of detail are obliterated. It is just a personal taste and each to their own.

When you have all the greyscale images click on the LR button in the tool bar and then make the iterations 3 and the curve radius as something like 1.3 to 1.7. My personal preference is a curve radius of 1.3. If the data will not support this sort of curve (ie if the processed image looks speckled) try using a wider curve until it gives a pleasing look to it. Trying different solutions will give you experience as to what works for your data sets. Remember of course that yoru equipment and the data collected is unique to your situation. There is not a series of settings that will work for everyone.

Once you have completed all the deconvolution for each greyscale frame, you need to name each one as to their correct colour. To do this, go to the Edit tab above the tool bar and click on Image title and then rename each frame with a simple r or g or b for the relevant colour.



Recombining Data (image below)

So the next thing you need to do is recombine all your separate colour channels into one RGB image. To do this click process and then click RGB combine. Or go to the tool bar and click on the red R button. When you do that you get the box appearing like that one below.

For those using colour AVI's you will not see much difference in the preview pane, however RGB imagers will find a large registration difference between the channels. That said both types of imaging can benefit from the recombine process.

To get all the channels registered one on top of the other you make changes in the plane shift boxes with postiive or negative numbers in pixel size. However before doing this ensure that each channel has the correct coloured channel. For example red should go in the Red Image, green to the Green Image and so on. Each time you make a change click the update button to see the changes.When you have the image registered correctly you can adjust the relative intensity of each channel. Once again use the update button to see how the changes have occurred.

Something to be aware of is how the registration works. If you are using RGB then this is extremely relevant. I use the features on the planet itself as my registration points. For instance on Jupiter I use the small storms and increase the magnification in the preview pane to look for faint red or blue marks and adjust each channel to get everything lined up. This means that the edge of the planet might have faint blue or red colouration where the planet has rotated while taking the data. You can remove this in Photoshop later if you like.



When you have finished aligning the preview panel look like that below. As you can see the planet looks well registered and the plane shift area has numbers of positive and negative values. These will change for each image that you do. I don't normally change the relative intensity figures unless there is a problem with one colour or another. I have found that exposing each channel at the same histogram point (or very near to it) will mean you don't need to make changes to the relative intensity.

When you are happy with your image, then click the ok button and the program will produce an image of your recombined data in its own window.



Gamma Correction (see image below)

Now you will need to reduce the gamma contained within the image. To do this click on scale and then click Gamma. The scale starts at 1 and you will need to reduce it to around 0.8. Now just click the ok button and the program produces another window with your updated image.


Unsharp Mask (image below)

Now you want apply some unsharp mask to sharpen the image just a little more. Click on Filter and then click on unsharp mask. The control panel for this will look like the one below. It is important not to go over board when using this filter. I like to use minimum blurring strength and a setting somewhere around 1.3-1.5. Click Ok when you have it set and see what the image looks like. It may take several attempts and adjustments to the power before you get it to where you like it. Try for subtle sharpening, don't be over zealous with your settings. The best images always show fine sharp detail.



When you click ok on the unsharp mask a new window will show the latest incarnation of your image. If you are happy with it save it in Tiff format and then move onto the next step.

PHOTOSHOP: FINISHING TOUCHES (see images and text below)

Photoshop is an image editing program with many powerful tools. For the purposes of this tutorial we are only going to deal with a few.

Once you start photoshop click on file and then click open. Select your image that you saved from Astra Image. When you have it open click on the Select Rectangular Marquee tool (dotted square) in the left side tool bar. Then click on the image and create a box of equal distances around the planet as indicated. Now click on image and click on crop. The image will be cropped to the box size as indicated with the Marquee tool.



Canvas Resize (see the image below)

Once cropping is completed I like to centre my image in a larger pane. This gives the perspective of placing the planet in space and centres the planet nicely for viewing.

First you will need to click the eye dropper and click very near the planet in the blackness of space. The will change the colour in the foreground colour and all you need to do is shift the background to foreground (the little boxes one slightly on top of another in the left hand tool bar). Now click Image in the top menu then go down to canvas size and click on that. The program will display the small box as shown below. The canvas size I use is 18.54cm x 13.9cm. Once you have your canvas size right, then click on OK.



Midtones Levels Adjustment (see image below)

The next thing you will need to do is adjust the midtones of the image. This increases the effect of the planet image looking like a sphere set in the void of space. So click on Image and then select Adjustments and then click on Levels. The levels control box appears. In the box you can see three arrows below the histogram. Click on the centre arrow and scroll down to around 0.96. Then click on OK.



Sharpening (see image below)

This is the final point of sharpening that I like to use. The aim here is to use smart sharpening to give that final crispness to your image. Click on Filter and then select Smart Sharpen. The box below will appear and I then select the slider for amount and use very low amounts until the image just starts to show some speckle. I have a very sharp LCD so I can get away with this a bit more than other monitors. If your monitor is not great, just back off the amount of sharpening a little once you get the speckle. Then click on OK.



Final Things To Do

The image is now completed, you can however add type to give details about your image and you name if you like. I will not bother to go into how that is done. The image below is the final image that I presented on 28 April 2008. It is not the best image I have ever produced but the data was good to use for this tutorial. I hope that you found it useful and I will review the tutorial as I work more on how to improve my images.


Back to How To