Better Mars Images with VirtualDub?

[Robert_T]

Hi All, cloudy again so decided to try some re-processing with VirtualDUb of a dissapointing Mars image taken through my Mewlon 180 last week.

I've been wondering for a while how much the distortion of the planet's disk is factored into the quality filtering in Registax - it seems to me to select or exclude frames on basis of feature contrast but not on the shape of the planet's disk. I decided to use Virtual Dub to scan through 1500 frames excluding those that looked distorted in shape and then processed the remainder to see it I could get a better image.

The original is at left below followed by two processings of the same AVI but Virtual Dubbed down to relatively distorted frames. Post processing in registax involved a small amount of LR deconvolution in Astra Image 2 with the final one having some despeckling applied in photopaint.

The latter images perhaps aren't as contrasty as the first, but I think the detail is better - a good indicator is the visibility of some cloud at the upper right that's invisible in the original - I'm guessing because feature alignment is better with the less distorted frames.

What do you guys think?

[venus]

Interesting pic of Mars there Robert, is that Olympus Mons I spy at 1 o'clock?

[asimov]

Hi Robert. Permission to play with your original?

[davidpretorius]

in virtual dub, did you hand select, ie frame by frame???

it is a real pain the way registax does not relegate badly out of shape images to the bottom of the pile just cos they have contrast

[Robert_T]

go for your life Asi - do your worst!!!;)




Hi Venus - my handy dandy mars maps says that Olympus Mons would be around about 4-5 oclock on the dial there.:) The level of detail here is unlikely to reveal it though sometimes a patch of orographic cloud shows where it should be .

cheers,

[Robert_T]

yeh, by hand frame by frame so it's not something you'd want to do routinely as it took nearly half an hour to go though 1500 frames - though Virtual Dub is quite easy to use so it goes quickly ....

[asimov]

Depends on one's interpretation as to weather my mucking about has improved upon your original. More details to be seen I reckon.?...At the expense of slightly more grain, but that's the trade off. Not to mention more of the dreaded bright limb coming thru.

[Robert_T]

Hmmm, you're right Asi there is a bit more detail in there and the grain added is only slight (better colour too) - what did you do, just an unsharp mask?

cheers,

[asimov]

If you look closely the blue cloud @ 2 o'clock is now present.

RGB split in picture publisher #8 stuffed around in blue & recombine. Unsharp mask/tone split-darkened red channel on midtones/sharpened in neatimage/overall colour balance tweak/despeckle.

[Robert_T]

oh ... and I thought it must have been something complex :rofl: :poke:

[iceman]

I often use VirtualDub to delete the bad frames, where the frame is obviously blurry due to bad seeing, bumping the scope or when the planet slightly drifts out of the FOV and I use my EQ platform speed control knob to bring it back in.

I then save the avi as a bmp sequence, which I then run through ppmcentre to reduce the height/width to 400/400 and centre the planet in the frame.

All of this appears to:
a) Help registax align on the features better
b) Helps registax process faster (thanks to reduced frame size)

It generally produces a better image for me, but sometimes I don't do it and I think I get just as good a result.

The reasons I don't do it all the time, is simply because it adds at least 20 minutes to processing an avi, much longer if you spend a lot of time deleting bad frames in virtual dub.

[matt]

Could someone post a link to a VirtualDub download page???

Thanks very much

[iceman]

http://www.virtualdub.org/

[matt]

Thanks again Mike:)

[Robert_T]

Hi Mike - that would be my experience here too; what I'm thinking (out aloud here in this thread) is that where Virtual Dub adds the most value is in deleting the badly shape distorted but still contrasty frames that Registax would have let slip through. That would suggest that the value of virtual dub is in pre-filtering avis's taken through very poor seeing, but perhaps not worth the effort otherwise. I think my Virtual Dub pre-filtered image (attached again below) certainly shows more detail than my original and perhaps even Asi's reprocess of my un-virtual dubbed original - or maybe it's just more noisy giving the illusion of detail? - though I guess this is a very subjective thing.

cheers,

[videoguy]

Robert

Well done mate. I agree, Virtual Dub is a great visual tool for discriminating (visually) the best images. In my opinion (and only mine) I think the pic at uppermost of this thread at right is best. Sorry Asimov..but in my opinion the natural look is lost in the reprocessed version. It is so important to maintain a natural reproduction despite the temptation to contrast the hell out of our images. I've certainly been guilty of doing this in the past but cringe when i view some of those results now.

Not wanting to blab on too much, but when i was on a 24" inch scope both recently and in 2001, I took the time to sit back with an eyepiece in place (rather than a CCD eye for a change) and tried to take in the view of this world from the perspective of early observers like Schiaparelli and Lowell - just for the heck of it. From this incredible perspective, one realises just how subtle the low albedo features are compared to brighter peach coloured desert regions and just how subtle the borders between dark and brighter are. Particularly during my 2001 run, I tried to reproduce the images to replicate (as close as possible) the colours my eye percieved versus the hues in which a camera (in some cases) more accurately replicates.

Bottom line: if the raw images aren't so great to start with, don't push em too hard in order to get more detail. Select and stack less frames of better seeing and enhance them lightly to create a more accurate representation of the view rather than introducing artefacts that spoil the picture and have no place in reality.