Help with very wide angle milkyway pic - completly messed up corners

[Jannis]

Sooo, longer focal lenghts are harder they say... Well i'm having more problems with my ultra wide angle then my 1000mm, lol...

Any tips on what i can do to correct the corners here? This is stack of 46x 20sec frames @ ISO3200 with 550d and EF-s 10-22mm@10mm. Taken in Fredrikstad, Norway, around midnight. Taken on a tripod, not an EQ mount. Stars look ok-ish at corners in single subs though, from what i could se on the camera.

Last attempt i thought it was due to clouds rolling in, but today it was cloud-free, and all frames look OK. I'm guessing maybe 10mm on an APS-C sensor might simply just be way too wide for DSS? Would using an EQ tripod help any when it comes to this?

This is stright out of DSS, stacked using "mosaic" cropping, to see what's going on around in the edges.

Full res JPG: https://www.dropbox.com/s/vsmk9wjikb33fm1/Milkyway%20test%202%2010mm.jpg

[baggywrinkle]

Have you tried correcting the corners in Photo shop.

It could be your lens making the disortion. But a nice image anyway.

[Jannis]

Thanks for the reply, but i can't really see no way that this can be corrected in post-processing, as the stars in both top right and bottom left corner are completly gome. Not washed out in light pollution or noise, but simplystacking error it seems.

I'm guessing this might be due to barrel distortion at 10mm that, when the starfield rotates, cann't be corrected in DSS? This would explain all the single subs turning out good, but not the corners after stacking.

I guess on a static tripod, if this is the cause of the issues, there's simply no way to take wide-angle short-ish exposures and stack them, like with longer focal lenghts? So i simply have to choose to either use a longer focal lenghts and do a mosaic, or use an EQ tripod - i'm i correct on this guessing?

[Oddsocks]

You shouldn't have that many problems stacking wide field images and the EQ tripod is more important for long focal lengths not short wide fields.

Star detection threshold setting and stacking mode is the most likely candidate for the stacking errors in your image but coma can also be a problem for DSS since it can no longer recognise star shapes accurately.

Try running the stack again with a change to the star threshold detection settings (advanced tab in "Register Settings").

Lower number = more stars detected but more easily fooled by hot or warm pixels and rich star fields containing mostly similar brightness stars.

Higher number = fewer stars detected, will stack using the brightest stars, less easily fooled by hot or warm pixels and rich star fields but may not stack accurately over a large field.

Experiment to find the best setting for a particular image set, this can vary from target to target and lens to lens and is not a constant.

Check also that hot/cold pixel detection thresholds are set appropriately so that DSS can tell the difference between a star and a hot/cold pixel.

The Mosaic setting is useful only for Mosaics, where the camera is moved between each single image or the images are obtained over a very long period, try stacking using standard mode.

[Jannis]

Thanks for the tips. Tryuing to restack now, with different settings. standard stacking (thought this only affected the framine though?), with star de3tect treshold at 30% instead of 10%, and with median noise filter on star detection. Stars went from about ~1500 to about ~130. Will see how it turns out in comparison.

[Jannis]

No lock, it actually only got 10x worse.. I looked over all the framed again, and found a few that shuoldn't have been in teh list though, as those were from when i was still framing the camera. I removed those, restacked, and restacked - but still no difference! >_<

this is what i get now. Like the original wasn't bad enough, lol. and a single frame...

Currently doing teh complete oposite now. set it very high, detecting about ~3500 stars at 5%, and no median filter on star detection, and will see what happends.

[Jannis]

I noticed it stacks fine if i only stack 10 frames at a time though, hmm... I guess i'll try that tomorrow, and then stack the resulting "stacks" and see hwo it turns out. Not supposed to be easy, eh?

[neil groves]

If it was easy Jannis, everyone and their Auntie would be doing it, frustrating as it is I am on my first DSS project, learning lots along the way, what keeps me strong though is seeing the images in the galleries from other people and knowing the results of perserverence will be very worth while.

Neil.

[ollypenrice]

You might try Registar for this kind of stacking. (Not Registax, note.) It corrects curvature from one image to match another. This is difficult for me to visualize but maybe your images have a moving curvature from one to the next as field rotation moves stars to different points on the lens distortion. Registar warps curvatures from different lenses to fit the chosen reference image. If Registar can't do it I think it probably can't be done.

I don't know how DSS works but making your defining image one from the middle of the run might help.

Olly

[Jannis]

Thanks for the tip olly. I've never tried Registar, but i'll look into it. Indeed the lens distortion and lack of tracking on an EQ mount is what i belive is the issue, as DSS doesn't appear to correct for this - thus letting me stack only a few at a time. But, let's say i stack 4 groups of 10 frames, then restack the 4 stacked frames - how would this affect the final picture when comparing to stacking all the frames at once? I guess it depends on the stacking method of course, but...

[Jannis]

It appears that registar worked Olly.

Hard to say how well though, as teh save button is disabled untill i've given them $149 for the full program >_<

But I know very little about the program though. It might be a good program, but concidering DSS is free, and have worked for all the images i've taken untill this one, i'm just not sure if it's worth it? Do you have any good/bad experience with it?

But, as for the main issue, wouldn't mounting the camera on the EQ mount fix this issue anyway?

As it's quite safe to say the cause is indeed field rotation and lens distortion like you said, and on an EQ mount with tracking, i'll still have the distortion, but no significant field rotation, right?

[Jannis]

Hmm, i completly forgot to try something that shuold be "common sense" i guess. Convert all teh raw frames to TIFF with lens distortion correction applied, lol. ;;

I noticed it didn't correct it all completly though (after taking a pic of a grid to correct manually, but even at 100% in canon DPP it wasn't perfectly corrected), but maybe it turns out a little better at least.

Might give ACR a try inste4ad of DPP if it still doesn't work, as i belive i can correct it even better there. Just completly forgot how to batch process there, lol.

[Jannis]

Well, did the stack with lens correction enabled - no real difference at all... -.- I guess using an EQ traipod is really only solution, or buy a software like registar that automatically corrects it.

Almost feels like i'm moving at warp speed when i look at it...

[dph1nm]

DSS really should not need more that 10 or so stars to stack, providing it is the same 10 stars! It only fits a rotation and a shift. Having several hundred is complete overkill and may even confuse the fit.

NigelM

[jezhughes]

Hey 

Lens correction in software will only work so hard so for you. The curvature on a 10mm gives a massively distorted edge and the flatter the field the better the blend. With large mosaics like the milky way, you need to take more shots closer together and overlap each, assuming that only the central 50% of frame will actually be stitch worthy (in my experience (through failure)). 

Also, i'd recommend using Photomerge in Photoshop (6) (File > Automate > Photomerge. There are various different options, one of which might work better with the data you already have.

You can also use lens correction under Photoshop's 'Automate'. Lightroom also has a lens correction facility, but i find it a bit useless..

Good luck!