DSLR Horizontal Banding...?

[AndyUK]

...Noel's actions...?

[MikeWilson]

Hi Mike - Yes, I'm quite depressed about this... . It's rather going to stop me in my tracks... :). (Have you seen the peltier finger mod - Hyper-Modification of Canon 450D (XSi) Cameras and here: Canon XSi 450D Peltier Cooler Modification | Rankinstudio - I thought filter removal / modification was scary, but that looks REALLY SCARY to me .

Hi Earl - I have to say, I'm REALLY impressed that you produced that out of a raw stack that I'd WELL over tweaked! Here's the image I did finally manage to get out of it without the banding coming through (much)

Goodness me. Those modifications are quite severe solutions to the problem of DSLR cooling.

I can't see myself doing that. Is there really no other alternative?

(I thought about attaching a fan to the tripod mount of the DSLR, but that would presumably cause vibration...)

[AndyUK]

Hi Mike - Alll I can say is, there is no way on this planet that I'm ever going to try anything like that(!), but the people who do have my everlasting respect! Although not as efficient, I suspect that the "box" will do a job, but again, it's not for me with my DIY skills...

[MikeWilson]

Hi Mike - Alll I can say is, there is no way on this planet that I'm ever going to try anything like that(!), but the people who do have my everlasting respect! Although not as efficient, I suspect that the "box" will do a job, but again, it's not for me with my DIY skills...

Is this modification even necessary?

With temperature matched dark files does the noise even matter?

I have a library of darks taken in the fridge that are quickly going to become out of date. I shall start making 'lukewarm' night darks and assuming that the chip heats at a fixed amount I'll record the ambient temperature of the lights and ensure that I have a set of matching darks taken at the same temperature/length/ISO to use. I will try to make at least ten for each temperature - gotta have something to do on a cloudy night!

Best,

Mike

[AndyUK]

I took a set of 20 x "fridge" darks which I assumed were all c. 4C. However, looking at the EXIF info, after the first 4 or 5, they ended up at 20C anyway. Maybe with enough stock darks and then reviewing the temp of all the subs it might be possible to match them up though(?).

Having said that, if the subs of an imaging run starts at (say) 12C and creeps to 20C and you use DSS, stock darks are going to create an averaged dark which will either under or over compensate...

Does this mean it would be best to temperature match each sub with a corresponding set of averaged darks at the same temp... and then you subtract a set of averaged flat's/flat darks or bias...? (Or is this making this way too complicated?)

[Cloudwatcher]

I don't know what causes the banding,has happened with my 450D on occasions,but there is a quick cure using Photoshop.

Make a separate layer(s) of the subject (stars,galaxy etc.)and having already chosen a background colour erase the original background leaving the subject against the new background. The edges may need a bit of blurring but the banding will have gone.

Heres an example using your image with,on the left,sections that have been treated as described.

[GrantEb]

I have also noticed a large variation in temperatures between subs, especially of late, as it has been warming up outside. I think averaged darks(taken at varied temps) will subtract "good" signal from your lights (that's what I think?)

I am busy working on a simple cooling solution for a DSLR, as I will need it when I get to warmer New Zealand. Will share it with all once it's working. (No cooler boxes etc!)

[MikeWilson]

I don't know what causes the banding,has happened with my 450D on occasions,but there is a quick cure using Photoshop.

Make a separate layer(s) of the subject (stars,galaxy etc.)and having already chosen a background colour erase the original background leaving the subject against the new background. The edges may need a bit of blurring but the banding will have gone.

Heres an example using your image with,on the left,sections that have been treated as described.

Wow, Cloudwatcher, you've really done a stellar job there.

Seeing your work alone has actually given me a little bit of hope, considering that I expect my images will turn out just like Andy's as the sub length and temperature increase.

Maybe with enough stock darks and then reviewing the temp of all the subs it might be possible to match them up though(?).

Andy,

With a great big sigh and a somewhat heavy imagers heart, I think you're probably right.

Perhaps the best approach then is on a cloudy night, leave a camera outside taking dark frames and then later inspect each one individually and catalogue it according to the EXIF temperature reading. Then, and only then can a fully comprehensive library of darks (with temperatures +1/-1 degree) be used in any given stack.

This is going to be a painstaking process. But at least it's something to do when it's cloudy!

Clear skies,

Mike

[MikeWilson]

I have also noticed a large variation in temperatures between subs, especially of late, as it has been warming up outside. I think averaged darks(taken at varied temps) will subtract "good" signal from your lights (that's what I think?)

I am busy working on a simple cooling solution for a DSLR, as I will need it when I get to warmer New Zealand. Will share it with all once it's working. (No cooler boxes etc!)

Oooh Grant,

You've got me all excited now! I only read your message after my last post on this thread.

I'm looking forward to seeing whatever you come up with

(It's not a "cold finger" is it?)

I expect Andy's DIY skills are better than mine! Shame there's no "aftermarket" accessory one can purchase to help alleviate the problem.

Best,

Mike

[dmahon]

Perhaps the best approach then is on a cloudy night, leave a camera outside taking dark frames and then later inspect each one individually and catalogue it according to the EXIF temperature reading. Then, and only then can a fully comprehensive library of darks (with temperatures +1/-1 degree) be used in any given stack.

This is going to be a painstaking process. But at least it's something to do when it's cloudy!

APT will kindly label all your frames with the ISO, duration and EXIF temperature in the filename - which makes it less painstaking. It's then easyt to build up a library of darks (just have a different directory for each) and also to match the right darks to your lights.

[AndyUK]

Cheers CW - That is quite amazing... I may well have to give that a go if I can't get shot of it - Thanks very much for the advice! From a purists perspective it does seem a bit draconian, but it's helluva lot better than seeing those bands!

Grant - Oh I DO hope you can find a solution - As Mike said, I think you could find quite a few people VERY interested in that...! You'll certainly have my undying gratitude... Please let us know if you come up with anything!!!

Not that I'm any expert, but I'm beginning to feel that the only real alternative is to subtract a stack of "matched" darks from each separate sub in separate DSS layers and then stack the whole lot together - Like you, I would also think that taking a single averaged dark from the entire stack would either remove too little or too much...

Mike -

I expect Andy's DIY skills are better than mine! Shame there's no "aftermarket" accessory one can purchase to help alleviate the problem

No way... Not possible! But I'd agree that if this is such a known problem, one might have thought that there was some enterprising person/company who would be happy to make / sell these, even if it was that Peltier cooled cooler box...?

What you've desribed is exactly what I had in mind - Leave the camera outside on cloudy nights for a few hours clicking away and then painstakingly sift through and sort them all by exposure and temp... And maybe then try the separate layer / sub in DSS (I wonder how many layers it caters for? )

[AndyUK]

APT will kindly label all your frames with the ISO, duration and EXIF temperature in the filename - which makes it less painstaking. It's then easyt to build up a library of darks (just have a different directory for each) and also to match the right darks to your lights.

That would make the job a LOT easier - Thanks - Good old Ivo!!!

[MikeWilson]

Thanks for the heads up on APT dmahon, I shall do that. Although the task is still painstaking and I'm not looking forward to it!

Hopefully the temperature during an imaging run doesn't change too much. I may make stacks based around +/- 1 degree centigrade. I know I can create at least five groups in DSS and I wish DSS had a temperature-EXIF based dark stacking mechanism.

So that means in addition to my current 20 BIAS, I need to collect about 20 FLATS and 20 temperature matched darks which need to run from 0 to 15 degrees at ambient temperature = 300 DARKS (or 2.4GB of data) for each sub length!

I'm thinking of collecting darks @ 60, 120, 300, 500 and 600 seconds. That should be 12GB of 1,500 dark images. Think that's enough?

[AndyUK]

Hi Mike - The chip temp difference that appears to have given me this result was 12C (6-18), but if you know that DSS will do at least 5, I guess temperature bracketting would be better than a single averaged stack(?).

Again, I'm thinking down the same lines as you... but hadn't done the maths. I suspect the trickier ones are going to be getting the lower temperatues as it would seem that as soon as the shutter fires, the chip starts to heat. So the lower temp longer exposure darks are probably going to be the ones with a LONG cooldown delay...

One thing's for sure though - With 5 temp matched DSS layers, my 4Gb RAM laptop ain't half going to chug, and as for drizzling, well...

[dmahon]

My imaging runs seem to be fairly constant over 4 hours the temp seems to change by perhaps 1 or 2 degrees. Even when it was zero outside most of the EXIF temps were 10. Now they're 15/16. Hopefully they won't be 30-40 in the summer!

But my SXVF-H9 arrived yesterday

[AndyUK]

My imaging runs seem to be fairly constant over 4 hours the temp seems to change by perhaps 1 or 2 degrees

Can I ask what exposure lengths and delay between exposures you're using to achieve that? As I noted, my 40D was taking 600s subs, started at 6C yet finished at 18C (over 22 subs) .

I'll have a look at the EXIF for all and plot a graph to see when it levels out - Obviously it will also be dependent upon ambient, which would originally have been 6C, but I don't know what it would have been at the end...

[Psychobilly]

Here you go Andy...

Astronomy Tools Action Set by ProDigital Software

[AndyUK]

Well, this is interesting, and I'm not sure what this means or how it helps in investigating where that banding's coming from, but I have a suspicion that wrong trees and barking might come into the sentence if we're purely looking at temperature as the possible cause.

The first sub appears to have been ambient (I guess the exif data is logged at the start of the exposure?), the 2nd frame was 15C, the 3rd 16C and all the others 18C.

For reference, the darks temps (despite being taken in the fridge) were all 20C apart from the first which was 15C...

(Any pearls of wisdom...?)

[MikeWilson]

How are you chaps reading the EXIF temps from your images? I was hoping Windows 7 would have a column in the file detail view for EXIF temperature, but it doesn't

I will have APT record the temperatures into the filename next time.

Andy - your temperatures are wild and it's interesting that your observations tend to suggest that the EXIF temperature is logged at the start of the exposure. This could be troublesome - perhaps it's worth compensating by pairing a light frame with a set of dark frames taken from +1 degree warmer temperatures given that EXIF temperature data may be unreliable?

@PsychoBilly: Noels actions are superb. I bought a license previously based on your recommendation.

@dmahon: Me = jealous!

Clear skies,

Mike

[AndyUK]

Hi Mike - I use KUSO Exif Viewer... Once installed, I've discovered that if you have all the raw images in one folder and left click on the first one, you can select the Exif viewer and it will display the exif data for that 1st frame (temp is 3 screens down), but you don't need to select each frame one-by-one in the folder - There's a right arrow that takes you to the next frame...

(Do check the actual temp of your fridge darks though - I once had a load of different exposure length folders under a folder with the title 4C as that was the temp of the fridge... They're now in separate temp folders, and the lower exposures are much colder than the longer ones...)

(Dmahon - SXVF-H9... I missed that... Yes... [ENVY!])

[joselin]

Could be a failure of a component above a certain temperature?

Would it be worth contacting the maker for advice?

[MikeWilson]

Could be a failure of a component above a certain temperature?

Would it be worth contacting the maker for advice?

Hi Joselin,

I think Canon would shrug this one off. Even the built-in settings only go as far as allowing 30 second exposures, anything higher than that is getting into a very niche field and noise is going to be expected - especially after 300 seconds and increasing with temperature.

Andy - good point regarding checking temperatures. I've done the same as you with the fridge darks. It might be a rather improper test but when I took the camera out of the fridge, it felt substantially colder than it does after an imaging run. It's just not that cold outside anymore.

I'll have to make some fresh darks, but getting a library together with seemingly random temperatures is going to take a while.

I'll start on the next cloudy but dry night, using APT and running the cables outside to the camera. I'll be ok as long as my cat doesn't take a wee on it.

Mike

[GrantEb]

Another option, which isn't a very good one, would be to use High ISO noise reduction switched on, and Long Exposure noise reduction switched on.

I know it will take a while to get lots of data though, but not to bad if youré only doing shorter exposures.

[GrantEb]

Have you guys seen this, to me it looks like the EXIF isn't actually the CMOS temp, but temp inside camera casing, or am I confused? The cold finger seperated the sensor in this case, but what about with no cold finger?

http://stargazerslounge.com/imaging-discussion/126136-dslr-sensor-heating-study.html

[MikeWilson]

Have you guys seen this, to me it looks like the EXIF isn't actually the CMOS temp, but temp inside camera casing, or am I confused? The cold finger seperated the sensor in this case, but what about with no cold finger?

http://stargazerslounge.com/imaging-discussion/126136-dslr-sensor-heating-study.html

I'll check that link shortly, Grant. I don't think it matters where the temperature is read from as long as it's consistent and gives a relative benchmark between light frames and dark frames. Match a light with a dark of the same length and temperature and you should be good to go.

Of course, our darks will probably be accurately recorded (in advance as the camera may already be cold) but the light frames may not have the right EXIF temperature if it's set at the start of the exposure and not the end.