7.53
#...
·
1
like
|
|---|
|
Hi all, One thing I’m having a hard time with is identifying “bad stuff.” For instance, after trial and error, I heard someone on YouTube say “watch your stars, don’t make circles in your stars.” If I hadn’t heard that I would’ve never noticed the circles! I wish we had a catalog of what is bad, like bad saturation, not enough contrast, too sharp, not sharp enough, bad background extraction, bad SNR, etc. It would definitely help noobs like me. Fortunately, I have a lot of AMAZING photos to study, to see good results. I’d just like more example of the bad stuff, too. Cheers, Rob |
11.36
#...
·
5
likes
|
|---|
|
Hi Rob, In my opinion, a catalog would be very difficult to make. Some of the aspects you mention are "to taste" of the owner of the image... In any case, here my thoughts on this: Bad saturation: For me, what is unacceptable is the point where an image is so saturated that one of the channels is clipped. I have seen this many times specially on strong blue reflection nebulas. Okey, the result is vivid blue, but it doesn't contain info, as is has been destroyed. Contrast: Almost the same, if we go too far with contrast, we will clip higlights, shadows...or both. Again this, for me, is unacceptable. Sharpness: This is not as influenced by clipping as the previous ones, maybe more a matter of taste. In my opinion, deconvolution it very dangerous because is performed just at the begining of the processing and then we put all the rest of the processing on top of this. This sometimes leads to clipping or noticeable artifacts. And (again, that's my "taste" ) deconvolution has a "signature" that I don't like. Gradients: I think is very easy to check it. Apply an temporaly extreme strecthing (like Boost STF in PixInsight or even more) and if you don't notice any gradient, you will not notice it when stretching properly. About this subject, I just tried a methodology called Multiscale Gradient Correction, released some time ago by the PTeam, that in my opinion is the best technique right now. I have had to deal with quite extreme gradients on an image of a dark nebula that I'm just going to release in short, and I have to say that without this method I would probably had to discard the whole project. https://pixinsight.com/tutorials/multiscale-gradient-correction/index.html SNR: This is a concept that I'm not sure if it has relevance once once starts the postprocessing of an image, I mean, you can have a very clean image, but if you stretch it too much, it becomes noisy... I'm far from being expert in this field, so maybe I'm wrong... It is interesting also to learn how to diagnose which process has been badly applied, or missing, from the images we see. Best Regards. |
|
7.53
#...
·
1
like
|
|---|
| @Alberto Ibañez Thanks for this info. This helps. I'll give Boost STF and MGC a try and keep your categorized notes handy. Cheers, Rob |
1.51
#...
·
1
like
|
|---|
|
The thing about badness is that you actually want a little of it or else your images will appear unnatural or lose detail. Rings around stars and uncorrected dust donuts are bad. Clipping — either over-brightening the signal or over-darkening the background — are not good because they lose detail ... but sometimes people are forced to choose between a noisy background and darkening to the point of clipping to try to suppress the noise. Noise itself has been described as the texture of the canvas your image is painted on: too much noise ruins the detail, but too little looks fake. Color, saturation, etc. , is largely personal taste but I find the biggest challenge is finding the right amount of noise reduction to enable you to stretch out as much detail as possible without over-stretching and tempting you to clip the background noise, but without appearing “plastic” and featureless by overdoing the noise reduction. Ie, finding the best texture for your canvas. |
