Here is the Eagle Crater to Endurance area:
Click to view attachment

(This was used to set the "parking lot" color level.)

-Mike

Some area South of Victoria. There are red artifacts on very flat patches at the top left side.
This is about the maximum size I can work with.

Click to view attachment

I saved a raw version here (7 MB), an high quality version here (10 MB) and some smaller resolution images here.

I did an edges map, but I find it more confusing than informative. At this area I don't see many long troughs, and they could be even dangerous to reach.

Shift Differential 10E5S Erebus Crater to Etched Terrain prediction compared with Ground Truth classification based on Oppy's images:
Click to view attachment

-Mike

Here is (I think) part of the putative route from Explorer Crater towards Endeavor.
This is from Tman's image higher resolution linked to in this post (this thread, post 123)

(This is the only chunk that I could find in jpeg format of the region beyond Victoria.)

Since the big dune wavelength was only 10 pixels (I assume this is at ca. 50% resolution) I used 5E2S for the differential shift.

Here are the results. (Edge effects were a pain, Explorer crater just beyond the NW corner had to get clipped)
Overlay with terrain (left) and Colorized only (right):
Click to view attachmentClick to view attachment

And getting back to Floyd's post, in the second "box" to the SE from Explorer crater (off screen, sorry!) there is a green zone surrounded by big scary dunes (in red). This central green zone could be another Erebus style fine dust deathtrap and should be examined carefully (or avoided).

-Mike

I remain blown away by the cleverly calculated maps posted by so many talented people here. I can't choose a favorite among them. It will be very fascinating to learn which were deemed the most helpful by the navigators.


...what climber said...

Hopefully, most here understand the rules you've made exceptionally clear from the beginning. I'm sure most of us are quite comfortable with those conditions.

My latest version of the erebus-to-edged traject for comparison, using the standard deviation / variance method:

Click to view attachment

With also the terrain south of Victoria using the same technique:

Click to view attachment

Furthermore I have updated my software tool, just for fun, with an option to create routes and the option to calculate total route distances, roughness along the route, and expected drive-durations (based on configurable settings for sol distances in each type of terrain). Although it is still very experimental, the software is now able to calculate its own routes through the terrain, in below picture the white and blue lines are computer-generated routes. Basically it will attempt to find the smoothest and fastest route through a ripple field, analyzing standard deviation and ripple-direction for the terrain ahead, this is nowhere yet accurate and it's definitely work in progress but it is fun to do and to watch the results.

Click to view attachment

Regards,

Geert.

I wonder if there's already a favored area in the official channel now where we'll roughly enter into the dune field.

Sorry for my late reply to this post, I didn't check over the weekend. I don't have time to download and convert stuff on my own, but if someone wants to do the conversion work, I will be very happy to host large files for all here to download and play with.

--Emily

I have played around with some more attributes. Previously, I was doing a 64x64 2dfft and finding the maximal sample in frequency space. From the full amplitude distribution in the 2d frequency space, I have computed several statistical quantities, and these are the most interesting ones:

wdist - the radial distance of the amplitude weighted average of the position, ie the centroid of the amplitude distribution (then converted to a physical distance by taking the reciprocal)
wstdev - the standard deviation of the amplitude distribution, which should be related to the width of the distribution - bigger values should mean more frequencies and thus more variation in ripple size within the window (no reciprocal here)
wangstddev - the standard deviation of the amplitude weighted angle distribution - bigger values should mean more angles in the window, which was my attempt to find the curvy ripples. This didn't seem to be successful.

I also replot the simple standard deviation of the values in the 64x64 window (no fft) in a color map consistent with the other three. I have scaled the color map in each case so that the data pretty well covers the whole color range.

I have also tried some classifications, with mixed results. The small ripples (eg in the center of erebus) are easily distinguishable, and the medium sized ripples on the east side of the crater are reasonably distinguishable. But I have not been able to highlight the curvy ripples, or reliably segment out the bedrock areas yet.

I think that Juramike's idea of looking to the ground truth to choose clasification examples is the right way to go, but of course Paolo is right that that job is best done by the guys with the most data. I would be happy to incorporate such guidance in the classifier.

I think I've done about all I can at this point without knowing more about what is helpful and what is not. But it has been a very interesting exercise. I was not optimistic at all in the beginning about the 2dffts, but they turned out to do a very nice job of distinguishing ripple spacing! Looking forward to the trip!

Bill

You know, Mike, you seemed to do a fine job picking out Purgatories by eye. Does there exist software that could take your dozen or two exemplars and search an entire HiRise image for 'close cousins'?

Scale shouldn't be a problem I wouldn't think, since we think Oppy is most vulnerable to something her own size. If absolute brightness -- or even brightness differences -- were too fussy a little hand-work could create variations, but I would hope the 'FBI fingerprint-matching software' tool I'm hoping is already out there has tweakable constraints.

WButler's approach almost does what I want but I would say he is looking for textures and I am talking about actual patterns.

All you folks are doing such fine work and creating such pretty and detailed and information-filled images -- but I find myself wondering what a simple gray-scale HiRise image decorated solely with red splotches everywhere a native Chinese speaker would recognize the character for Purgatory-like fish-scales (if you know what I mean), I find myself wondering what that would look like.

Very good job Bill!



That is indeed exactly what I'm doing with my own tool, basically it should tell you something about the roughness of the terrain, a lot depends however on how big you set the window space and the sample rate within this window. It sure seems to mark out the bigger ripples and the more rough terrain but it is difficult to note which terrain will give most problems to oppy.



See my earlier attempts at finding ripple angles via the same method, If you take a very big window (a big distribution of ripples) it averages out nicely on the average ripple-direction, however it is very hard to get correct directions for each and every ripple as this method very easily gets confused and to pick out each and every ripple correctly you need to set the window quite small and that doesn't give you a big enough distribution for the tool to work on, at least that seems to happen to me.

What I hope will give you an indication of 'cross-ripples' is taking the standard deviation in the frequency range of the ripples, so instead of the amplitude distribution, you try to find an average wavelength for the ripples across the window, then take the standard deviation from the frequency spectrum, a wide spectrum would mean a great variance in different wavelength, and so a bigger chance of cross-ripples and chaotic terrain... As yet I have not been able to do this as it takes a lot of computer-power to do this accurately.



Correct, I run against exactly the same problem. Bedrock I can find if I set a small window (small distribution) as then its variation doesn't match the surrounding ripple-fields and it will get classified as 'flat', however if you set an average 64X64 window the bedrock areas are just too small to get noted. The dangerous curvy-ripples are still alluding me...

In an other post it was mentioned to try some 'fingerprint' trick, in other words pick out a sample of a 'dangerous' ripple, then compare this across the image to spot the dangerous ones. I'm going to try to do this, but I've never written FBI-software so I don't know how long it will take to get that trick working ;-). As is often the case, the basic idea is quite simple, however the technical implementation is a different story, you'll have to take care of image-scale, different viewing-angles, ripple-directions, brightness, etc, etc, and then compute more of less a figure which tells you for which percentage the sample corresponds to the 'dangerous' ripple... It sure looks nice on TV, finding a face or a fingerprint like that, and I'm sure it can be done, but my time is limited and this is something I've never tried before, still it's fun to do...




I also think at a certain moment you more or less reach the limits of what you can get out of these HiRISE images by computer-power, it's fun to try the 'fingerprint' thing but by the time I've worked that out probably anyone who looks at the pictures which his/her own eyes has already spotted all the 'dangerous' ones. Exactly the same can be said about 'dust-traps' and such, without multi-spectral images I have not the faintest idea how to spot these using software although they are 'obvious' to anyone looking at a picture...

As has been said often before, in the end it's the human eyes, and the experience of the drivers, looking at the pictures which are required to find a route, it's a lot of fun to see how far you get with software-processing but with only the HiRISE images there's just no way you can ever get anywhere close to the true experience of the drivers and their eyes.

I think we more or less reached the end of software-processing, and thanks to all the tremendous work in comparing results with ground-pictures we now know more or less what each trick will tell us. As soon as the 'new' HiRISE images arrive we can loosen our tools on them, but most of all we will need people with experienced eyes, just looking at the pictures and simply marking out the 'dangerous' spots.

hi all,

here is a 'Purgatory' animated gif that demos a version 0.2 of the DEM model I have been working on.

Green is drivable
Dark is not

Purgatory is slightly left and south of center (bright in the HiRise / dark in the DEM model)

Comment: most of these dunes gullies are drivable, just stay away from dark 'sides' of obvious dunes....navcam and planners are doing an awesome job!
<annoying diagonal is still present...arrrgv[x] !)

Click to view attachment

Cheers

Hi Geert,

I may not have been as clear as I wanted about the attributes I was computing. The only attribute I posted earlier today that was computed in the image domain (eg actual pixel values of the image) was the one I called stddev, which was the standard deviation of the pixel values. All the others were computed in the frequency domain. I took a 2dfft of a 64x64 square of pixels, keeping the lowest 32x32 amplitudes of the frequency components, and take their absolute value. So I think your comment:

"What I hope will give you an indication of 'cross-ripples' is taking the standard deviation in the frequency range of the ripples..."

is what I'm already doing. That is, to compute wdist I find the centroid of the frequency samples, or the amplitude weighted mean of all of them: mean_fx=sum(amp*fx)/sum(amp), and mean_fy=sum(amp*fy)/sum(amp) over all the samples. This gives me a coordinate (mean_fx,mean_fy). The radial distance to this point is the mean frequency of the ripples, and its reciprocal is the mean wavelength, which is what I call wdist. I started out just using the maximal amplitude sample for that coordinate, but I thought this might be too influenced by noise (I called this dist earlier). Interestingly, I found that dist was significantly more than wdist (like 2x-3x), for reasons I still don't understand, but the trend of the two quantities is similar. wstddev is the standard deviation of this distribution, or the width in frequency space of the ripple distribution. I don't take the reciprocal here, because a large frequency distribution width should correspond to a large wavelength distribution width. There is probably a more accurate way to do it, but I think to just get the trend right this is fine.
Similarly, wangstddev is computed in the frequency domain. Each cell in the 32x32 fxy space is assigned its polar angular value, and then I compute mean angle wang=sum(amp*angle)/sum(amp) over all the cells. wangstddev is computed similarly, but using the standard deviation equation rather than the mean. Since a single angle in frequency space also maps to a single angle in image space, I expect that wangstddev should measure the spread of angles present in the actual image. I haven't played with different sized ffts, and this may be helpful, but I feel like most of my pictures are rather splotchy, and I don't have a good enough feel for the ground truth to know which ones are the most helpful.
I also think a fingerprint algorithm may be helpful here, but I don't have one handy

If I have completely misunderstood your post, I apologize. Carry on and have fun!

Bill

Thanks for this explanation, then you are doing indeed exactly what I was thinking about! Presently my own tool uses only the stddev calculation and I'm basically estimating ripple directions by taking a stddev calculation of the brightness along a line with a certain range, and then turning this line 360 degrees round centered on the specific pixel (more of less a 'radar-screen'). In the direction of the ripples you can expect the largest stddev as in this direction the brightness of the image will show the biggest differences. This is far from perfect but gives a reasonable value as long as you take the 'range' big enough, however if you decrease range in order to work on a smaller scale, the noise will grow and the accuracy will drop dramatically, so it doesn't work for individual ripples.

In my software tool the window is standard on 64X64 pixels, but you can set it is high as 256X256 or as small as 16X16. Setting the window too high will result in missing out on the 'small patches', setting the size too low results in too much noise.




I think we need to take in mind that these ripples are far from sinusoid, to me they look more like waves in very shallow water. From my own textbooks about shallow-water waves I know that under those circumstances a lot of 'normal' formula's no longer work on them, and this might also be a handicap with the whole fft calculations. I do not presume to really understand much about the dynamics of these types of ripples and then it is very hard to interpreted exactly what we are 'measuring'. Also our 'data' is extremely limited to only the HiRISE imagery, that's why we are very rapidly approaching a situation where you simply run against the limits of your data-set...



Neither have I ;-). Never done something like that and it's way beyond my normal line of work and I'm afraid that re-inventing the wheel takes some time... I have a basic idea on how to proceed with the software but that's extremely processor-intensive and I'm worried that in the end my poor laptop will be calculating for one week just to find one ripple which everyone else spots in 10 seconds by eye

OK, I think it is time. Please post or e-mail me your best and final effort that classifies:

- Purgatory style ripples (red color)
- Large ripples (blue color)
- everything else (sand, outcrop, small ripples) (green color)

These are the only classes I need. The more classes you submit the harder it will be for me to analyze the image. If you believe your algorithm is better at one class only, generate a map that uses the appropriate class color and then leave the rest as green.

Please use the MAP PROJECTED HiRISE PSP_001414_1780.

You do not need to process the entire image, just the portion south of Victoria. If you are submitting an image, i prefer uncompressed PNG, PNM, TIFF, but if you can't safely convert into any of these formats, just send whatever format you can. If you send a presentation, please send it as PDF. When you submit your map, please make sure you are keeping a copy of the procedure and parameters you have used to generate the map. Once we get the new HiRISE near Endeavour I will ask you to generate a similar map and want to make sure they will be consistent.

Thank you.

Paolo

I can't put time into this myself, so apologies in advance for suggesting something I can't do myself, but it occurs to me that everyone seems to be trying to solve the relatively-difficult problem of finding ALL the "safe" routes when it's only necessary to find a single one.

To me, this looks like a plausible application for a perceptron net (aka "neural net"), where the input features are more or less the same features you guys have been using already, but for a single square centered on the current location of Opportunity (to start with), and the output is scores for different possible drive directions (perhaps just eight possible ones). You rule out directions that would backtrack, and then select the best score from the remaining ones.

To train the net, use the data collected in the drive so far, with a 10% hold-out set.

With luck, that will trace out a decent path from the current location to the target, and it should definitely go right down the middle of ripples. Where it could fail is if it wanders into a dead-end, but there don't seem to be a whole lot of those; you could handle that problem manually.


Wish I had time to try this myself . . .


--Greg

Ooohh... do we hear the sound of Oppy gunning her engine impatiently..?

I would disagree. We need to see the big picture so that we can make choices on route, hop between good areas if necessary. Infact, Paolo has specifically state that he doesn't want a single route drawn out.

My turn to desagree !
What I understand is that Paolo need "only" a map little South of Oppy's actual position at this time.
So, I come back to the proposition I made at the beginning of this topic : let's give Paolo just this and then, when we, more or less ,understand where Oppy's going, let's produce the next similar map. Will make sure to be only one or two weeks ahead of Oppy.
This do NOT means that we do not need the whole picture. This means we need a very accurate map "only" a bit ahead of Oppy and this will become more efficient with the HiRISE image.
This is what I understand ok, with personal interpretation for sure...

Any chance somebody can be a true Hero and chunk out a full resolution jpeg of this area?

-Mike

We do. The big picture dictates the driving trend. There's no point having a 'two week' map only to find we're two weeks down a dead end into a dune field.

This
http://maps.google.co.uk/?ie=UTF8&ll=5...041113&z=15
doesn't tell me how to get to London.


This, however, does
http://maps.google.co.uk/?ie=UTF8&ll=5....262451&z=8

I'll see what I can do - it will be pretty large though.

Maybe you are misinterpreting this quote from Paolo's post:
"You do not need to process the entire image, just the portion south of Victoria."
(There's a difference between 'just the portion South' and 'the portion just South'.)

He's saying don't bother with anything North of Victoria, but process as far south of Victoria as the image goes. He even says he will provide MRO coverage even further south as soon as it is available. He gives categories of areas he wants marked. No mention of marking paths, or one path, is made. It seems he is reserving this duty for the MER team. Imagine!

I don't know if any of you have read Edward de Bono's "The Six Thinking Hats," but what Paolo is asking for is White Hat Thinking: just the facts, without our interpretations thrown in. The problem is coming up that we are all trying to be "useful" (or even worse, "right") by injecting our own egos and assessments into the request. That's not what is needed at this point.

He's not asking for us to solve the problem, just to lay out the data so that we get all the facts on the table. That's the way we can be most useful at present.

I highly recommend de Bono's process for anyone who has complex problems to solve. See his website for more info about his system.

Awesome!!!

Would it be easier to break it up into two or three pieces?
(I suppose they could always be reassembled later into a lower-res global picture)

-Mike

Let me make sure my intent is clear. At the moment I want to map the hazards. Since there are many algorthms that give good answers, I want to see how difficult is to map a path using these tools you have created. There's nothing particularly magic about finding a path but it requires knowledge of a few things that I am not sure I can talk about.
Also for the moment there is no need to analyze the area north of Victoria (terrain we have already driven on).

Paolo

Hi Paolo,

Could you give us some guidance on what you mean by large ripples? Are they what is in the eastern half of Erebus, or do you mean even larger than that? Approximately what percentage of the full Erebus image we have been playing with would you consider large ripples? Thanks!

Bill

Oh yes, I break the jp2 into chunks 5152 pixels high so you'll get it in those size chunks - but 29794 x 5152 is still a HUGE jpg.



I'm at that stage now and am assuming:

Green - no ripples or very (Insignificantly) small.
Blue - Ripples smaller than Purgatory (or maybe even the ripples around Purgatory if not enough comes out red)
Red - Larger than Blue ripples

I'll let you know when its done.

Among the things you can talk about, what is the most unusual, unexpected or bizarre navigation cue you used for mapping a path?


Paolo Amoroso

I consider large ripples those that are higher than purgatory and with distance between ridges that were wide enough for navigation. 3-5m between ridges?
Remember that what is characteristic of purgatory style ripples, is not necessarily the size, but the fact that the ripple is curved.

Paolo

Click to view attachment

Bill,
I like what you're doing with a 2D FFT.

Since Paolo mentioned he's interested in finding spots where the ripples converge (as happened in purgatory) it strikes me that it would be possible to do some editing in the frequency domain and suppress the main trend of the periodic ripples, and then convert back into the spatial domain. In such a reconstructed image, the main trend of the ripples (generally north - south in this region around Victoria) should be suppressed, while other irregularities (crater rims, ripples intersecting the main trend) would remain. Those would then highlight areas to avoid -- or at least where the Rover Drivers should exercise caution.

Do you (or anyone else) think you could give that idea a try?

All the best, Steve M

The tie-down cleats on the wheels. The rovers were secured on the lander by bolts and the location on the wheel has deep cleats that have the characteristics of capturing soil on on side of the and depositing the soil on the other side and they leave pretty distinct marks on the terrain. We can measure the distance of these markings and determine slip even without running Visual Odometry. When the slip is very large (80% and up) the cleats leave the soil in small mounds. In Purgatory we were able to determine pretty accurately the actual motion of the rover during sol 446 (the sol where the rover got embedded). This gave us some information to verify our physical simulation in the sandbox. That means we wanted to make sure the simulated embedding we generated in the sandbox was as accurate as possible. This allowed to verify the way Opportunity was going to recover from the embedding event. Not only we wanted to make sure Opportunity would be reovered, but also wanted to compare the progress on Mars vs the progress in the testbed

Paolo

If you are on Linux, you can make crops of the image using IAS Viewer and a xorg.conf tweak.

Open /etc/X11/xorg.conf and under the Screen section, subsection Display, add a line "Virtual 2560 5120" (or any other dimensions), save and restart. The desktop will cover a bigger area, which you can navigate moving the mouse to the screen edges. Run IAS Viewer maximized to save the crops. If the algorithm 'bleeds' at image edges you'll need to keep some overlapping margin.

Steve M.,

I think that filtering in the 2dfft domain is a nice idea. I think it will be important to not just cut out some components, but smoothly suppress them with some sort of dampening field. Sharp edges in one domain makes ringing in the other. I've got a couple of other ideas to work on first, but I'll keep it in mind. Thanks!

Bill

OK Paolo here is what my Fourier transform analysis technique throws out.



Three versions- raw gray-scale, the 255 gray-scale levels turned into colour green->blue->red and a version trying to reduce it down to just three terrain types. No or very small ripples = Green, Ripples = Blue, Large Ripples of Purgatory size or greater = Red.

Yes, I know this is not exactly what Paolo wants, Red = exclusively curved Purgatory type ripples but I think that red areas both contain the majority of these and also are the areas where the ripples are so large that maneuvering around 'Purgatory types' would be difficult, and may well cause just as many problems as 'Purgatories' when trying to move across them in any direction but due South.

I think this gives a good overview, in particular picking out several 'rays' outward from Victoria with smaller ripples compared to the surroundings.

James

Now to convert those JP2's to JPG's for Juramike - if anyone else wants them, PM me.

Here is the analysis applied to the full (and lower ) resolution HiRISE image over the region just south of Victoria.

This shows what the algorithm does and doesn't detect, and the kind of terrain each colour represents.



James

James I am in awe, seriously.

I'm thrilled just to be lurking on this thread.

Does anyone else see an interesting 3D effect on James' maps? On my screen (maybe because of my colorblindness?), the red areas seem to be floating above the blue. Cool effect! Awesome work!

Is there a Nobel Prize for ripple analysis?

Edited: Seriously, James should be invited to be a Rover Driver for a Day.

James! Stop reading my mind!

Paolo

Woah! Very very impressive!
The first thing I noticed is the appearance of these "rays". Could this be the result of subtle geological features (created by the impact of the object that created VC) hidden by the ripples? It looks like this algorithm shows a little more than the smoothness of the terrain. Kind of validation of the model.

pac

I have tried that already Greg, this is basically how the 'route finding' option in my software tool works and it's only a small step from my daily line of work. Basically you create a software 'rover', move it in a certain direction, then analysis the terrain at the new position, and 'clone' the rover into a couple of hundred new 'rovers' which each of them set out again from the new position. The terrain analysis predicts for each rover how far it would get into this direction during a certain time frame. Then from each new rover-position once again 'clone' each rover and repeat the whole process. Then basically you wait and see which rovers arrive at the preset destination and you analyze the 'history' of each rover arrived at the destination to find the route it has taken (offcourse the first rover to arrive can be expected to have taken the fastest route, but there might be other requirements). This is a very basic description, there is a lot more to it (software rovers are given some own 'intelligence' in selecting routes, there is a method to remove 'stuck' rovers and to trace history for each rover, etc, etc) but the overall process is quite simple, at least for a computer (it has to 'manage' several thousand 'rovers' at the same time). A lot of route-planning software you use daily is working via this method.

Big problem with this whole process is that it works nicely here on earth if I guide a ship across a stormy ocean, where I have access to large databases with all information I need and all performance data of the ships, etc, etc. However, for oppy we have only the HiRISE images and they just don't provide enough good information to do this type of work (you have to make a tremendous amount of assumptions...). It does give you a route (see my earlier post) but as there is no method to check all the assumptions you made, the accuracy will be very low ("GIGO").

Quite apart from that, as has been mentioned already, the question asked was not to 'find a route' but just to provide a map-analysis of the terrain. This doesn't mean that we can't trail beyond the requested task and suggest a route (this is a public forum) but as I already stated you have to take into account that we simply don't have the data so any suggestion would be guesswork at the best and most probably completely useless to the real drivers. I'm very happy to leave finding an actual route to the professional rover-drivers then we can stay in the backseat

I agree with pac56- we are clearly seeing impact-related "rays" that are far less obvious when looking at the unprocessed image. The impact left its mark in ways that are still influencing the ripple distribution, despite considerable erosion since then. I wonder what the mechanism is. Subtle topography? Variations in the degree of bedrock fracturing?

John.

A lot of it reminds me of putting a stone (or whatever object) on a windy and sandy beach, then watch the results after some weeks or months, often you won't be able to see the original object anymore, but the sand-ripples at this position will surely be disturbed and often in an area far exceeding the size of the original object. Once there is the slightest disturbance in the sand, the whole ripple-creating mechanism will enhance this I guess. If you look closely at the ripple-fields on the images you see a lot of area's where the ripples clearly are 'disturbed' both in direction, wavelength and height, although it is (to me) often not obvious what has created this disturbance but there might be something deep below under the sand.

Which makes me wonder if there is a way of more or less 'dating' the various craters within a ripple field, based on the amount of disturbance they have created in the ripples...

First slice:

Click to view attachment

Processed at 1:2 resolution and color coded at 1:4. I didn't mark Purgatory-like dunes nor attempted to substract featureless flat areas, that needs extra steps at full resolution. I'll try to do that later with a better computer.

I like the green streams between blue dunes.

My own version of the same image, using standard deviation in pixel brightness. I find more 'blue' and less 'red' however this purely depends on how you define ripple-sizes. The 'rays' are slightly visible but less obvious using this method.

Click to view attachment

Anyone who likes the full-scale version, just send me a pm and I'll send it by email.

Geert.

Differential shift method 10E5S of the first slice (I stuck with the green-yellow-orange-red color scheme:
Click to view attachment

-Mike

(My goodness, we've all been busy tonight, haven't we! )

It's better to use the red-blue-green gradient.

Later color components can be substracted from different models and combined together, i.e. green from Fourier to override wrong blue areas from edge density.

Crap.

I already colorized the rest of them. Oh well, at least I saved an interim version for each.

-Mike