20 replies to this topic

[teb76]

Hello all, I was reading this article on SPCC on Pixinsight that says

 

 

when working with OSC cameras is to always use drizzle. Besides the fact that only a drizzle integration can provide optimal results by avoiding interpolation, some de-Bayering algorithms may modify color proportions at small scales, where interpolation of missing color data in CFA patterns plays an important role.

 

My setup is Celestron C8 XLT + Touptek ATR294C + Optolong L-Quad enhance. Being in the city center, the seeing is not that good so if I correctly made my check I should almost always be in oversampling.

I understood that in case of oversampling, drizzling has the only effect of increase the noise so it should be avoided, neverthelesse, because of the a.m. article and some other youtube video I decided to give it a try, with the following result on NGC 2903:

 

here is the No Drizzled integration

here is the Drizzled integration

 

I'll not share for now what's my opinion but I'd like to have your opinion about these two images, if you see one better than other and why. This would help me in understanding how to look to an image and what to consider to judge an image better than another. 

Thanks

[afd33]

I preferred the drizzle version for the most part. After opening them, to me the drizzled version looked slightly better. The stars in the corners were ever so slightly more round and I thought the galaxy itself looked slightly better. I then ran BlurX on both of them with the same settings. I used automatic for nonstellar sharpening.The stars were nice and round in both. I would say the galaxy itself looked pretty much the same in each, but I would lean just a little towards the non-drizzled version. Then I ran NoiseX on both, .5 denoise and .25 detail. I didn't want to over do the denoising, but I wanted to still apply a pretty good amount of it. With the noise that's left, I prefer the look of the drizzled version. It's hard to describe, it just looks a bit more smooth I think is the best way to put it.

 

Overall, I'd vote that I like the drizzled version better

[markalot63]

Hello all, I was reading this article on SPCC on Pixinsight that says

 

 

My setup is Celestron C8 XLT + Touptek ATR294C + Optolong L-Quad enhance. Being in the city center, the seeing is not that good so if I correctly made my check I should almost always be in oversampling.

I understood that in case of oversampling, drizzling has the only effect of increase the noise so it should be avoided, neverthelesse, because of the a.m. article and some other youtube video I decided to give it a try, with the following result on NGC 2903:

 

here is the No Drizzled integration

here is the Drizzled integration

 

I'll not share for now what's my opinion but I'd like to have your opinion about these two images, if you see one better than other and why. This would help me in understanding how to look to an image and what to consider to judge an image better than another. 

Thanks

 

Non drizzled on the left, drizzled on the right. 

 

BlurX used in correct only mode

 

BlurX non stellar .5

 

BlurX settings used:

 

I think the drizzled is better, more contrast.  It looks different enough I find myself wondering if this is the same image.  Clearly it IS.  

[scanner97]

I'll confess I didn't look at your images - because I've already looked at a lot of comparisons.  For me personally, there are times when I can hardly tell any difference, and there are times when the drizzling looks better.  It seems to depend on a bunch of other factors.  So, is drizzling a "best practice"?  Sure.  Is it necessary to get a decent image?  Not.  Will it always give an image that looks better to me.  Not.

 

In the same way, calibration frames are a best practice.  But, if you have a modern sensor, good optics that match the sensor, and no dust on your filter, you may not notice a difference.

 

Or should you guide just because you're shooting a DSO and not a planet?  It depends...although certainly guiding is a best practice for DSOs.

 

Always worth asking whether the results justify the effort, and the answer is always personal and subjective!

[scanner97]

I think the drizzled is better, more contrast.  It looks different enough I find myself wondering if this is the same image.  Clearly it IS.  

I would certianly agree.  I've seen other comparisons where the difference was not nearly that apparent.  I'm curious about your gear and seeing, i.e. whether you were over- or under-sampled.  I would guess under.  If it's over, that will be a new (and interesting) data point for me. 

 

Thanks for sharing!

[markalot63]

I would certianly agree.  I've seen other comparisons where the difference was not nearly that apparent.  I'm curious about your gear and seeing, i.e. whether you were over- or under-sampled.  I would guess under.  If it's over, that will be a new (and interesting) data point for me. 

 

Thanks for sharing!

These are the OP's images, in case I confused people with my post.  

[teb76]

Non drizzled on the left, drizzled on the right. 

 

BlurX used in correct only mode

 

BlurX non stellar .5

 

BlurX settings used:

 

I think the drizzled is better, more contrast.  It looks different enough I find myself wondering if this is the same image.  Clearly it IS.  

I did the same kind of comparisons and to me the non-drizzled was better by far! Now I'm curious and I'm going to repeat the comparisons, perhaps I was a little drunk at that moment 

[scanner97]

Never drizzle drunk!  

 

(And the other processing makes a difference...)

[archiebald]

Hello all, I was reading this article on SPCC on Pixinsight that says

 

 

My setup is Celestron C8 XLT + Touptek ATR294C + Optolong L-Quad enhance. Being in the city center, the seeing is not that good so if I correctly made my check I should almost always be in oversampling.

I understood that in case of oversampling, drizzling has the only effect of increase the noise so it should be avoided, neverthelesse, because of the a.m. article and some other youtube video I decided to give it a try, with the following result on NGC 2903:

 

here is the No Drizzled integration

here is the Drizzled integration

 

I'll not share for now what's my opinion but I'd like to have your opinion about these two images, if you see one better than other and why. This would help me in understanding how to look to an image and what to consider to judge an image better than another. 

Thanks

 

I also didn't look at the images.  I experimented with drizzling last year but found the following, which was backed up by conversations with others;

• It only really works with high numbers of subs and integration time
• It absolutely requires subs to be dithered (although you are probably doing that anyway)
• It often leads to increased noise
• At best, you might see some small improvement, more commonly almost no difference.
• Depending on sensor size and PC specs, processing time can become incredibly laborious and slow.

So I gave up with it, found it to be a complete waste of time for most of my images.

[teb76]

BlurXterminator full + SPCC to me non-drizzled continue to look sharper and with better colors.. I'm suspecting I do something wrong compared with your pictures above

 

[markalot63]

BlurXterminator full + SPCC to me non-drizzled continue to look sharper and with better colors.. I'm suspecting I do something wrong compared with your pictures above

 

Immagine 2025-03-27 233042.jpg

For the drizzled picture I plate solved then spectrum photometric color blah blah calibration to fix the colors.  

[teb76]

For the drizzled picture I plate solved then spectrum photometric color blah blah calibration to fix the colors.  

That's what I did too

[Yomamma]

On anything that is not undersampled I drizzle at 1 in PI and I like the results

[WadeH237]

The suggestion to always drizzle with OSC cameras is not about resolution.

 

The reason to drizzle is so that you can use DrizzleIntegration.  The benefit here is that it improves color accuracy at very small scales.  The reason for this, is that drizzle integration can avoid interpolation of the individual color channels.

 

You should even do this with an image that's over sampled.  You would just set drizzle so that  it doesn't increase the horizontal and vertical pixel counts.

 

There is a prerequisite for this:  You must do an effective dither when you capture your subs, otherwise drizzle won't have enough sub-pixel information.

[teb76]

The suggestion to always drizzle with OSC cameras is not about resolution.

 

The reason to drizzle is so that you can use DrizzleIntegration.  The benefit here is that it improves color accuracy at very small scales.  The reason for this, is that drizzle integration can avoid interpolation of the individual color channels.

 

You should even do this with an image that's over sampled.  You would just set drizzle so that  it doesn't increase the horizontal and vertical pixel counts.

 

There is a prerequisite for this:  You must do an effective dither when you capture your subs, otherwise drizzle won't have enough sub-pixel information.

Yes I always dither, at least every 3 minutes. I got the point, I can use drizzle 1 instead of drizzle 2 if I'm oversampling. Thanks for the explanation

[archiebald]

The suggestion to always drizzle with OSC cameras is not about resolution.

 

The reason to drizzle is so that you can use DrizzleIntegration.  The benefit here is that it improves color accuracy at very small scales.  The reason for this, is that drizzle integration can avoid interpolation of the individual color channels.

 

You should even do this with an image that's over sampled.  You would just set drizzle so that  it doesn't increase the horizontal and vertical pixel counts.

 

There is a prerequisite for this:  You must do an effective dither when you capture your subs, otherwise drizzle won't have enough sub-pixel information.

This is something new to me - must remember to give it a try sometime. - thanks
 

[JF1960]

On anything that is not undersampled I drizzle at 1 in PI and I like the results

How does a 1X Drizzle help with undersampling?  As compared to 2x?

[WadeH237]

How does a 1X Drizzle help with undersampling?  As compared to 2x?

1x drizzle is not for handling under sampled data.  The reason that you would use it is to get the color accuracy improvement that I mentioned above with a OSC camera.

 

I don't know of any reason off the top of my head to use 1x drizzle on well sampled mono data.  At that point, you could just skip drizzle altogether.

[teb76]

The suggestion to always drizzle with OSC cameras is not about resolution.

 

The reason to drizzle is so that you can use DrizzleIntegration.  The benefit here is that it improves color accuracy at very small scales.  The reason for this, is that drizzle integration can avoid interpolation of the individual color channels.

 

You should even do this with an image that's over sampled.  You would just set drizzle so that  it doesn't increase the horizontal and vertical pixel counts.

 

There is a prerequisite for this:  You must do an effective dither when you capture your subs, otherwise drizzle won't have enough sub-pixel information.

That's indeed what I understood. I used drizzle at 1x but as I shown in my pictures below, colors of non drizzled integration look more natural to me. I don't know if I'm doing something wrong

[Yomamma]

How does a 1X Drizzle help with undersampling?  As compared to 2x?

I drizzle 1x for oversampled and in the range sampling.  I use 2x  for undersampled data 

[tectonik]

Fred Vanner posted some good comments on drizzling with OSCs (pixinsight forum)

 

https://pixinsight.c...th-a-osc.20669/

 

"Just remember:

    Drizzle only works with well-dithered data. In this context "well-dithered" means that the sub-pixel remainder of the dither displacement must perform a good random sample of the pixel - in both x and y axes.
    Drizzle is a statistical process; it gets better the more frames you use, but requires a reasonable minimum number of frames to work properly - so you typically need more frames if you want to see any improvement.
    The more oversampled your image, the less improvement you will get from drizzle. If you are much more than x2 oversampled you will see virtually no difference in the drizzled images. This is because the spread of the PSF has just the same effect as debayering with symmetrical (bilinear) interpolation; never use the default VNG interpolation - it is the worst possible choice for astrophotography (though the difference is small)."

 

Tek