hiepphotog wrote:
Fred, I think this action would benefit with an extra step of color noise reduction. That would turn most of the red/blue hot pixels into monochromatic pixel, then dust&scratches would work better.
This action is for hot/stuck pixel removal only. It does not touch color/luma noise.
I advice removing luma/color first and then run that action as a final step in Photoshop.
If you use LR, I get good results with color noise all the way to "100" and detail to "0".
You know I posted that 78 second raw file that would be great to play with. It was shot with LE NR off. The link is in the thread titled long exposure. .
Be a nice idea everyone used the same file make there adjustments than repost it. I'm okay with that
Next on my hit list. I had dark frame built into my phase backs and certainly cleaned up extremely well. For some Astro work folks shoot many shorter images than stack them. Also a lot of this cleans up in post a great deal. I view this as not a big deal. Actually having a hard time finding any real fault with this cam. Some thing you avoid. But the good news is pretty much everything that came to light can be fixed fairly easy with firmware. Trick is getting past the hysterical posts get down to the real facts and get that data in Sonys hands to work on. Give this a couple weeks and this will settle down too.
juvx wrote:
My questions is why not just simply keep the LENR ON.. i mean that's why its there, to combat this.... Has anyone tested this with ON vs OFF?
Because LENR limits the raw output to a mere 12-bit file. I think there is a desire to wrok around that limitation. Disappointing that these extra measures are needed.
juvx wrote:
My questions is why not just simply keep the LENR ON.. i mean that's why its there, to combat this.... Has anyone tested this with ON vs OFF?
Plus LENR will take more time. Correct me if I'm wrong but LENR takes another image of equal time as your first image. If you're shooting 30 seconds or more, doubling that time will mean change of light and missing the shot.
yunjo wrote:
Plus LENR will take more time. Correct me if I'm wrong but LENR takes another image of equal time as your first image. If you're shooting 30 seconds or more, doubling that time will mean change of light and missing the shot.
juvx wrote:
My questions is why not just simply keep the LENR ON.. i mean that's why its there, to combat this.... Has anyone tested this with ON vs OFF?
I have done the comparison in the past but no longer have the pictures. LENR is very aggressive that I see no point of using it. For critical image, I would rather do spot removal myself; that shows how bad the in-cam LENR is. But like most astrophotographer, you can take all the night pictures you want then when you are on your way back, you can take dark frames for these pesky hot pixels.
snapsy wrote:
Btw, here are 8-minute exposures on the 5DSR @ ISO 100 with no LENR. I presume the 5DS/R got the same long-exposure noise improvements as the 7DM2 that Roger mentioned in his article.
I've been using Canon bodies for long exposure night photography for a long time, virtually always with LENR. I'm quite impressed by those samples with no LENR! (I have the 5DsR, but I have not done any long-exposure night photography with it yet.)
I wonder if the more visible hot pixels, or rather their visibility, might be a downside of the otherwise positive ability of the Sony sensor to register lower light levels/higher DR? (The law of unintended consequences at work?)
And, yes, all camera systems — even very good ones — have their strengths and weaknesses. I have yet to see the camera that is perfect in all ways.
gdanmitchell wrote:
I wonder if the more visible hot pixels, or rather their visibility, might be a downside of the otherwise positive ability of the Sony sensor to register lower light levels/higher DR? (The law of unintended consequences at work?)
It's a good question and I only guess at the answer. Previous generations of the Sony Exmor sensors have a propensity for purple, amp-type noise in the deep shadows near the noise floor for normal exposure times at High ISO and it's usually worst at one edge of the frame. It's been speculated that the noise is from the dense collection of on-chip parallel ADCs clustered at the edge of the chip where the noise emanates from, the same on-chip ADCs that are attributed with the Exmor's excellent DR. Perhaps those are also the source of some of the dark current leading to the hot pixels. Sony also might not have made dark current noise management a priority, unlike Canon which appears to have made some technical advances in this area, at least as reported by Roger Clark (see the dark current section of his 7DM2 review here; the 5DS/R presumably inherits these advances since the sensors are similar in many other regards).
I have looked at the shots i took yesterday, and to complete my test i did some with LENR enabled as well. On my shots i did @ ISO 800 to keep under 30s exposure, the confetti like noise pixels were visible, but just (LENR off). After bumping the exposure 1.5 stop they were pretty noticable over the entire frame. The shot i did with LENR on removed all the confetti noise and left a cleaner image to work with. But, like others said before, LENR is not always an (good) option.
Correct me if I'm wrong (because I am often wrong), but doesn't increasing ISO (say to 1600) increase power to the sensor, thereby, increasing amp noise and also introducing hot pixels at a shorter shutter speed because power has increased to the sensor, allowing the sensor to heat up faster than say at base ISO of 100?
yunjo wrote:
Correct me if I'm wrong (because I am often wrong), but doesn't increasing ISO (say to 1600) increase power to the sensor, thereby, increasing amp noise and also introducing hot pixels at a shorter shutter speed because power has increased to the sensor, allowing the sensor to heat up faster than say at base ISO of 100?
Lowering ISO for this shot would have caused >30s exposure which would let the raw drop to 12bits leaving less shadow detail (according to earlier posts from Fred).
therealthings wrote:
I have looked at the shots i took yesterday, and to complete my test i did some with LENR enabled as well. On my shots i did @ ISO 800 to keep under 30s exposure, the confetti like noise pixels were visible, but just (LENR off). After bumping the exposure 1.5 stop they were pretty noticable over the entire frame. The shot i did with LENR on removed all the confetti noise and left a cleaner image to work with. But, like others said before, LENR is not always an (good) option.
Leave LENR "On" when shooting at ISO 400 or higher. The higher read noise will dither any degradation caused by the 1-bit reduction.
Whenever you are limited by time to use LERN (astrophography) you could take one or several dark frames after your shooting and remove hot pixels in Photoshop later on OR just run the action I created. It will remove all hot pixels from your image while leaving sharpness and noise alone. https://www.fredmiranda.com/forum/topic/1380165/2#13144133
Fred Miranda wrote:
Leave LENR "On" when shooting at ISO 400 or higher. The higher read noise will dither any degradation caused by the 1-bit reduction.
Whenever you are limited by time to use LERN (astrophography) you could take one or several dark frames after your shooting and remove hot pixels in Photoshop later on OR just run the action I created. It will remove all hot pixels from your image while leaving sharpness and noise alone. https://www.fredmiranda.com/forum/topic/1380165/2#13144133
Fred:
I virtually always do use LENR for my long exposure night photography. (Most often of urban subjects and in the range of perhaps 1-20 minutes exposure time.) I have heard of the method you describe — making a single dark frame exposure and using it to subtract "noise" from the other frames.
Since you have used it, what can you tell me about its consistency by comparison to the in-camera LENR method. The theory is that the noise changes as the ambient temperate and sensor temperature very, but perhaps (as with so many things) you have found that this theory really has no significant practical effect on photographs and therefore the single dark frame image works virtually as well as generating a new one for each exposure?
(Yes, I could test it myself, but if you already have... ;-)
Fred Miranda wrote:
Leave LENR "On" when shooting at ISO 400 or higher. The higher read noise will dither any degradation caused by the 1-bit reduction.
Whenever you are limited by time to use LERN (astrophography) you could take one or several dark frames after your shooting and remove hot pixels in Photoshop later on OR just run the action I created. It will remove all hot pixels from your image while leaving sharpness and noise alone. https://www.fredmiranda.com/forum/topic/1380165/2#13144133
Thanks for the clarification. The ISO 100 shots look clear to me.
Fred Miranda wrote:
Leave LENR "On" when shooting at ISO 400 or higher. The higher read noise will dither any degradation caused by the 1-bit reduction.
Whenever you are limited by time to use LERN (astrophography) you could take one or several dark frames after your shooting and remove hot pixels in Photoshop later on OR just run the action I created. It will remove all hot pixels from your image while leaving sharpness and noise alone. https://www.fredmiranda.com/forum/topic/1380165/2#13144133
Fred, if I'm reading your comments correctly, then you get the same results with LENR off + running the action, as you would if you took a dark frame?
What I'm also wondering about is a noise comparison between the 5Dsr and the A7rII, and A7r. It seems that the A7rII creates more hotpixels than the other cameras. But, let's assume we can mitigate the hot pixels with either LERN or your action...in that instance, how do the overall noise levels compare?
I virtually always do use LENR for my long exposure night photography. (Most often of urban subjects and in the range of perhaps 1-20 minutes exposure time.) I have heard of the method you describe — making a single dark frame exposure and using it to subtract "noise" from the other frames.
Since you have used it, what can you tell me about its consistency by comparison to the in-camera LENR method. The theory is that the noise changes as the ambient temperate and sensor temperature very, but perhaps (as with so many things) you have found that this theory really has no significant practical effect on photographs and therefore the single dark frame image works virtually as well as generating a new one for each exposure?
(Yes, I could test it myself, but if you already have... ;-)
Dan, for your needs, since you are not restricted by time, I would just use LENR. It's convenient and gets the job done. Taking a "single" dark frame manually and subtract the noise should not be noticeable better.
It's only better if you average the noise by taking more than one dark frame. For astro, I usually take up to 15 dark frames shooting at the same exposure and temperature as my light files. I do it at the end of the shooting session when it's convenient and the sensor temperature is stabilized. The dark frames will be averaged later on to create a master dark. The reason for this is that you when average images together you reduce the amount of noise (by the square root of the signal).
You can't use LENR when shooting star trails because it doubles the exposure time creating gaps between your shots.
The reason for A7 series shooters to take a black frame manually is to avoid the reduction in precision to 12 bits. However, we are probably already using Bulb mode and higher ISO, so it probably won't make any difference in the final image anyways.
Fred Miranda wrote:
Dan, for your needs, since you are not restricted by time, I would just use LENR. It's convenient and gets the job done. Taking a "single" dark frame manually and subtract the noise should not be noticeable better.
It's only better if you average the noise by taking more than one dark frame. For astro, I usually take up to 15 dark frames shooting at the same exposure and temperature as my light files. I do it at the end of the shooting session when it's convenient and the sensor temperature is stabilized. The dark frames will be averaged later on to create a master dark. The reason for this is that you when average images together you reduce the amount of noise (by the square root of the signal).
You can't use LENR when shooting star trails because it doubles the exposure time creating gaps between your shots.
The reason for A7 series shooters to take a black frame manually is to avoid the reduction in precision to 12 bits. However, we are probably already using Bulb mode and higher ISO, so it probably won't make any difference in the final image anyways. ...Show more →
artd wrote:
Fred, if I'm reading your comments correctly, then you get the same results with LENR off + running the action, as you would if you took a dark frame?
What I'm also wondering about is a noise comparison between the 5Dsr and the A7rII, and A7r. It seems that the A7rII creates more hotpixels than the other cameras. But, let's assume we can mitigate the hot pixels with either LERN or your action...in that instance, how do the overall noise levels compare?
Yes, that action does a pretty good job and you should try it for those occasions when LENR is not possible. (Shooting star trails for example)Remember to remove chroma noise in LR before running the action in PS.
