# 21 Feb, 2018 18:39
Roberto ColombariI think it is a good idea to write those objective criteria guidelines. All submitters, reviewers and judges should consider examining every image in full resolution detail. I use to spend 30-60 minutes everyday as a submitter to inspect carefully the IOTD candidate images. In addition to the list above, I should add:
5 - Out of focus stars in the center part of the image (or at frames' borders in mosaics) due to poor focus/guiding;
6 - Strong dark halos or other deconvolution artifacts around stars;
7 - Bleached background due to excessive noise reduction;
# 21 Feb, 2018 20:52
As a submitter, here is the way I look at images to make a choice :
1/ First selection on "aesthetic" base : do I like the general look of the image, the composition, originality , colours, general balance, level of detail considering the diameter of the optics, the signature or watermark on the image (size/color/place)…?
2/ Originality of the image : related to a hot event (comet, transit, eclipse, etc.), or addition of special data (mix of wide and narrowband for example, etc.) ?
3/ Technical appreciation (assuming all technical datas are filled), including :
- no visible issues with pre-processing : artefacts, doubled stars, bad correction of flats ;
- no visible issues with acquisition (drift, elongated stars due to bad tracking, etc.) ;
- quality of mosaic or animation if appropriate;
- gestion of noise (not to noisy, but not over-denoised) ;
- aspect of the background ;
- aspect and colors of stars + halos + quality of star reduction/deconvolution if relevant ;
- dynamic of the main objects ;
- gestion of colors and transitions ;
- saturation of colors ;
- quality of general contrast (and local contrat improvement if appropriate) ;
- quality of the mix with narrowband if appropriate;
- size of the image (highly reduced or cropped or not).
The technical appreciation is made in coherence with :
- the setup and the total exposure time : a bad image has in every case no chance to be selected even if made with a little setup, but it will be the same for a "just in average" image made with a high-class setup (many k$) and long exposure time (+10h). But a good image made with a little setup or not so many hours of integration will have a chance despite some little technical issues which do not compromise the global visual impression of the image.
- the difficulty of the object (size, magnitude..).
- the technique used for Imaging (long time exposure, Lucky Imaging, etc.).
4/ In case of equal value between many images, I usually favor the less common object or objects that have not been awarded in the last days.
And sometimes, I just have an instant crush for one image !
For Professional data, I am obviously more demanding about processing… I will even say that no fault (or what I consider to be) is tolerated !
The elements I never take into account for selection :
- the author ;
- the number of likes of the image ;
- the number of IOTD/TP of the author (global or recently) ;
- the category (deep sky/planetary/pro-data/remote or not/terrestrial..)
I hope I have not forgotten too many elements.
PS : I forgot one thing : excepted special cases, I do not choose 2 or 3 pics of the same object on the same day.
PS2 : one HUGE omission : I do selection on a well calibrated pc screen !
# 21 Feb, 2018 21:01
I agree more or less with everything you wrote.
It could be a nice starting point for the guidelines!
# 21 Feb, 2018 21:42
Hahaha. J-B, maybe you should just be the sole IOTD/TP selector. We can all donate a coffee a week to have you spend all day on Abin making selections.|
Seriously though, I’m in agreement pretty much across the board.
# 21 Feb, 2018 21:53
And be sole responsible of all injustices in selection, no thanks !
The list of critera may seems very long when written like this, but in real life you can quickly make a correct judgment on many of these points when you look the image in real size. In fact, it took me a lot longer to write this message than to make a daily selection !
Obviously, the elements I pay attention to are the same I pay attention to when processing my own images, so it's kind of an habit… I imagine that a lot of submitters/reviewers do the same.
# 21 Feb, 2018 22:04
Roberto ColombariThanks Roberto.
I didn't talk too much about pro-data processing, because I'm not familiar with this practice, so I surely not have the good reflexes or relevant elements in head.
I have decide to make a try with pro-data processing, in order to have a better idea of the difficulties it implies (and maybe stop beeing a little obtuse about it) but for the moment I have difficulties to get the calibration files, so I am far from being able to get an idea of the challenges…
# 21 Feb, 2018 22:38
|We can chat about it, if you are interested!|
# 22 Feb, 2018 15:44
I just wanted to add in here that I don’t think downsampling from the original image scale should be a disqualifier. After all, we all use different focal lengths and cameras. Take an example case where two images were taken through the same optics, one with a camera that has pixels half the size of the other. Is the fact that the image from the camera with smaller pixels gets reduced to 50% resolution a disqualifier? I should think not. It’s not much different from binning upon capture.|
Besides, I think the spirit is to not judge the optical quality and resolution of the image unless it impacts the aesthetics. In the cases where it obviously lacks resolution and that detracts from the image, or the resolution is a standout feature of the image, I have no issue judging on that quality. But, I have some issue with a blanket criteria on how you present images that can be considered for IOTD.
….just an opinion…
# 22 Feb, 2018 16:34
thanks for your PoV.
What is meant is a little bit different.
Think about a STF8300 (indipendently from the scope). The image size as it comes out from the CCD (b1) is around 3326 x 2504 pixels.
If you process this image and put on astrobin resampled at 1300 x 1001 pixels (40% of the original size), IMHO, it shouldn't be considered equally to others at full (or nearly) res. for an IOTD.
Hope to have clarified the topic.
# 22 Feb, 2018 17:00
Roberto ColombariWhat if I used the camera on an exceptionally long focal length scope? Or decided to bin it 2x2 for some reason?
Haha. I’m not trying to be difficult or otherwise a nuisance, but there are cases where downsampling would be perfectly acceptable in my opinion. I suppose I don’t find downsampling an issue unless it detracts from the aesthetic quality of the image. Of course, then it becomes less quantitative and subject to more opinions. I’ll leave it here, because we may just respectfully disagree and I’m not involved in the judging anyway!
# 22 Feb, 2018 21:40
I just wanted to note that I found this whole post to be an excellent way to look at selecting top picks and this line in particular is a good point, and encouraging to me (and I'm sure other beginners). For example, my image of Simeis 147 was selected as a Top Pick today. I am happy to receive this honor, but I am very aware that it has some technical issues with the stars, but I think considering it was shot with a camera lens this is understandable, and as Jean-Baptiste says above, does not "compromise the global visual impression of the image". I do NOT think my image is worthy of IOTD, but I think any Top Picks guidelines should give a little leeway for the judges to consider the setup that produced the image. IOTD on the other hand should really be the best of the Top Picks, both technically and in visual impression. My 2 cents as a relative newcomer (started 2 years ago).
# 22 Feb, 2018 22:09
Not really what I would call a "beginner's image" in my opinion !
I remember I saw it for selection (I'm not going to say if I selected it or not ! ) and I noticed the setup and the 18h of integration… it well deserve a TP !
# 22 Feb, 2018 22:34
Don't underestimate yourself!
# 22 Feb, 2018 23:27
The way JB explained how he picks his selection sounds great. We should also remember that judges are selecting based on what is available them. So setting standards saying the image has to pass a certain criteria may not work. They may not have an image for that day that meets the requirements. We should give them the benefit of the doubt. They are doing a good job.|
The only thing I would change about the process right now is to require a minimum amount of information if one wants to be eligible for IOTD so the judges can make a call. The scope, camera, overall integration time could be a start.
I am not sure I agree with resolution requirement. A judge can make the call if the image shows enough detail or not. In this image I purposely cropped to show detail without even going to full screen. The resolution is 1266x934 but it's still 0.7" per pixel. The overall resolution is not the only factor. The plate solve function didn't work on it so it doesn't show this data.
I also think Nico's image is well deserved for TP. You are not supposed to pixel peep on widefield images.
# 22 Feb, 2018 23:33
If you crop you don't lose resolution and that is completely fine.|
What I was talking about is a drastic resizing (except for huge mosaics and other few cases).
If you have a 3.3k x 2.5k pixels image (example: using an STF8300) and you upload it at 35% of its size is probably because you want to "hide" some "issues" like low SNR or artifacts. In this case I really prefer to pick images that are uploaded at full res.
# 22 Feb, 2018 23:43
If the plate solve fails, how are you going to know what the imager did unless you go create an FOV and figure it out. Processing an image IS "hiding issues". When we do noise reduction, we are hiding the noise. When we reduce halos, we are "hiding" them. It should be up to the imager to represent the image the way they want and it should be up to a judge to decide if the imager did a good job on hiding the issues or even know anything is hidden.
# 22 Feb, 2018 23:53
I understood the intention of Roberto's proposal to curb the reduction of image size, meaning to avoid resolution loss beyond the optically useful compression threshold (the case presented by Tolga is interesting in this respect). In the particular case of huge files, one may include a link to it. (What is the purpose of assembling a giant mosaic if nobody could see it in full detail?) Memory and bandwidth are becoming cheaper everyday.|
# 22 Feb, 2018 23:53
In this case you won't decide based on the resolution. Simple as that. The cases where astrometry fails anyway are not so many.
BTW, it's not a mandatory rule. It must be up to the submitter/reviewer/judge to distinguish case by case.
I recently saw a couple of top picks where the images were cleary reduced about 30%-40%. IMHO, these ones shouldn't get to TPs.
TolgaWith all the due respect to everyone's processing procedures, noise reduction and all the other processing algorithms are far different that taking the image and upload it as postage stamp.
# 23 Feb, 2018 00:01
ive created some gigapixel sized mosaics, uploading those as full resolution is certainly not practical. Even when binned 2x2 or 3x3 they can be damned huge.
# 23 Feb, 2018 00:06
AtmosFearICYes, I understand! If I want to download them, how can I do so? I would love to see them!
# 23 Feb, 2018 00:15
AtmosFearICI took one of the image I am processing and I resized it just for testing purposes up to 20.000x13500 pixels. I saved it as JPG in PS quality 9 (which actually excellent).
The size is 61Mb. I don't think it's too huge, isn't it?
# 23 Feb, 2018 00:17
Your assertion that the reason why people bin was they were hiding something. Now you are building a strawman argument by misrepresenting what I am saying as people should be allowed to upload postage stamp image. Have you ever heard of a KAF-3200 chip? It's 3.2 MP with 6.8um pixels. No matter what you do the image will be the size of an postage stamp. This image (not mine) is size of a postage stamp but yet it's one of the best I have seen. Noise reduction IS hiding noise. You are in fact hiding the noise.
All I am saying is to leave it up to the judge to decide how much hiding they accept.
# 23 Feb, 2018 00:23
We still don't understand each other or it's me that I am not using the correct words.|
If KAF-3200 has a 3.2MP sensor, no problem with it. Your image will be 2184px ∙ 1472px. I am comfortable with that!
What for me is an alert is if the sensor is 16MP and you upload the image resized (I am not meaning cropped) 1500px ∙ 1500px.
# 23 Feb, 2018 00:31
Roberto ColombariRoberto I understand what you are saying. I am trying to show you that there are other reasons why people may post images that look smaller. I know someone who use a 695 chip (QSI690) with a C14. They bin the camera during imaging. They don't resize during processing. I am just saying there are other reasons why image can be smaller than what we think it should be besides they were "hiding something". Wde should not simply disqualify it. A judge can look at it and decide.
# 23 Feb, 2018 00:35
TolgaI agree on that and that is why I wrote just few posts above:
|You have no new notifications.|