(Auto) ETTR (Exposure to the Right): -- History & Beginners Guide

[pjburnhill]

Excellent, thank you Alex & dmilligan, very helpful!

For recording HDMI to ProRes, would the same principle apply? Or does the ML digital ISO not affect HDMI (I assume it does)?

[Quentin]

Does Auto ETTR work real time in Video Mode ?
I dont think so because its a hard procedure that takes time, hard to capture and store at the same time.

When I record video, which is the last valid calculation ?
When I press the Button ?

Thanks for the illumination

[Audionut]

It works fine in real time, in video mode.  You just may not like the exposure jumps.

A good option would be to ETTR just as you would do in RAW (to maximize the SNR of the input data), then use ML digital ISO (the one from image fine-tuning) to get something close to your final look.

Never thought of this.    

And BTW, this thread in 4th on a google search for ETTR, it's probably time I cleaned this sticky up a little.


edit:  Oh, I could have sworn it worked the last time I tried it (about 12 months ago  )

[a1ex]

Right now it stops adjusting as soon as you press record - to avoid the exposure jumps.

It can be made to work while recording. However, it will require a deflickering algorithm when extracting the DNGs, or a fast (real-time) algorithm to tweak ML digital ISO on the fly for H.264.

There is real-time access to Canon's YUV histogram data, which could be used for this purpose.

[Quentin]

I hope there will be switch for this.
Sometimes you know that you will shoot and the lighting is steady and predictable so dynamic ETTR is not essential.
Sometimes you have a complex shot with cast changes in lighting conditions where a dynamic ETTR is necessary.

Thanks in advance for the Magic!

[Audionut]

Big thanks to RenatoPhoto for the original OP, however it was probably a little long in the tooth, and more importantly, outdated.

I have edited down to the specific information, and updated it for the latest changes in the module.  Let me know if I have missed something, or made a mistake.

[SalvadorPadeny]

Very good thread.
I have been a huge supporter of ETTR during the years as it's another way of making the most out of our cameras.
@audionut: sorry for breaking the thread flow but apparently (from the electronic cigarettes package i see) you are from Australia.
What area?[/color]

Definitely it was pretty nice thread so from which region he belonged?

[mrd777]

I have a question. If I use zebras for exposure, technically, can't I achieve the same result and ETTR myself?

Thank you,
Mr D

[Audionut]

You can use any of the raw based exposure guides and ETTR manually.

[budafilms]

@Audionut thank!  (From the wrong black Level thread).
I have read all the thread and I will to trust in ETTR!

It's a wonderful tool!

[tetsusaiga]

I've been using ML intervalometer to create regular time-lapses  (i.e., not day-to-night time-lapse or vice versa; just plain old vanilla ones).  After discovering and watching a few time-lapses created by other ML users, I found that they used AutoETTR and ML other features to create their time-lapse.  I'm striving to reach their level of expertise and I've been diligently reading the following articles to understand AutoETTR:

http://wiki.magiclantern.fm/ettr
http://www.magiclantern.fm/forum/index.php?topic=5693.0
http://www.luminous-landscape.com/tutorials/expose-right.shtml
http://www.luminous-landscape.com/tutorials/optimizing_exposure.shtml

These articles have been very helpful in helping me (kind of) understand AutoETTR.  But all the technical talk has made it very confusing and difficult for a beginner or intermediate DSLR user that has no background in computer science, physics, optics, digital signal processing, etc., (like myself) to fully understand what AutoETTR is, what each ETTR parameter means, how to adjust each ETTR parameter to achieve the desired result, etc.  I feel like I need an advance degree just to keep up with the discussion. 

So I'm politely asking the ML community and DEVs, please, can someone explain AutoETTR to me in layman terms.  I have a Canon 600D and my main focus is to use AutoETTR to shoot time-lapses; particularly, holy-grail time-lapses.  I would very much like to create a time-lapse similar to these guys: 

Here I go with my very basic understanding and questions.

(1) Tigger Mode:  From what I gathered through my reading, and for my particular interest (time-lapse), the "Trigger Mode" should be set to "Always On."  This feature essentially means that the camera will always analyze  its surroundings to adjust the ISO and shutter speed for the next shot.  In other words, the camera will take a picture, analyze its surroundings after taking the picture, adjust the ISO and shutter based on the analysis, and take another picture using the new adjustments.  These steps are repeated for each shot.  Is my understanding correct?  And is the camera analyzing its surroundings to adjust the ISO and shutter, or is it analyzing the last picture taken to adjust the ISO and shutter?

(2) Slowest Shutter:  As the parameter implies, this is where you adjust your slowest shutter speed.  But I'm not completely sure why you need to set your slowest shutter for AutoETTR to use.  So, for example, I'm shooting a day-to-night time-lapse.  The shutter speed during the day shots are at 1/1000.  As the sun goes down, the ISO and shutter speed would be automatically adjusted by AutoETTR to compensate for the decrease in light.  The shutter speed eventually reaches 1/16 (i.e., the speed I set for the "Slowest Shutter" parameter).  What would reaching my set limit do to AutoETTR?  What significance does "Slowest Shutter" have?  What does increasing or decreasing the shutter speed do in terms of AutoETTR?  What is the slowest shutter speed recommended for day-to-night time-lapses?  Can someone place explain this in layman terms.

(3) Exposure Target:  The description of this parameter on the first page of this thread says "Where to place the highlights with respect to overexposure."  I'm not really sure what that means.  For example, I have highlights that are overexposed, such as the sun.  My "Exposure Target" is set to -1EV.  Now what?  Where is the highlight "placed" with respect to overexposure?  Where does increasing or decreasing the EV place the highlights with respect to overexposure?  What EV is recommended for day-to-night time-lapses?

(4) Highlight Ignore:  From my understanding, this parameter is used to ignore a highlight in your frame, right?  So if I'm taking a picture of the sun, "Highlight Ignore" will ignore the sun?  Can someone please explain this and how it can be used for day-to-night time-lapses.

My brain is fried from reading all day and trying to understand all this before asking.  So that's all the questions I have for today.  Any help would be great.  Thanks.

[Audionut]

It's important to first have a good understanding of what ETTR is, what it's benefits are, how it works.  It's probably a good idea to read some stuff about histograms also, so you understand what a histogram represents.

1.  You're understanding is correct.  In live view modes, it always has data to analyse, when shooting through the viewfinder, it only has the data from the last image taken.

2.  Shutter is motion control, pure and simple.  Since it also controls exposure, ideally you probably want the shutter to remain open for as long as possible, subject to motion.  ISO is post exposure gain.  It's a little hard to explain without technobabble, but suffice to say, ISO is not an exposure control!  ML knows this, so it will open the shutter for longer periods, subject to the slowest shutter setting, before increasing ISO.

3.  Exposure target is where to place the highlights.  It's probably best set at 0.5 EV, for maximum ETTR, until you have a good understanding of the theory, and the operation of AutoETTR.

4.  I think of highlight ignore as an specular highlight control.  The value set here describes the percentage of pixels in the entire image, allowed to be overexposed (or more correctly, above the limits set via AutoETTR).  So for instance, if you're shooting a scene with a specular reflection, this control allows you to compensate for that reflection, so that AutoETTR will ignore it, and ETTR the important areas of the scene.  Since this is based on a percentage, it's value depends highly on the scene being shot.

[tetsusaiga]

Thanks milligan for responding to my PM.  Audionut, thank you for responding to my questions and trying to dumb it down so that I could understand, although I am still confused.  Please bear with my ignorance, I'm really trying my best to understand the camera's sensor, the ML modules, and how to use these modules to create a sunset/sunrise holy grail timelapse.

To answer milligan's PM question, yes I read the thread:   http://www.magiclantern.fm/forum/index.php?topic=5705.0.  I also read this thread:  http://www.magiclantern.fm/forum/index.php?topic=5693.0, as well as a number of other threads about ETTR.  But after reading these threads, I was even more confused.  I think it's because of my lack of understanding of how the camera's sensor works and the technical details of photography.  I really want to learn and strive to be an expert in timelapse photography, so please help me achieve this goal.

Just to give you an idea of how I shoot my regular timelapses (i.e., not sunset/sunrise timelapses).  First, I set my ISO to 100 for daytime timelapses.  Then I set my shutter speed to what I believe will give me the right amount of motion blur so that the final timelapse is not too choppy or have a staccato effect.  I take a test picture and see how much motion blur is in the picture.  If I'm happy with the amount of motion blur, I set the ML intervalometer to a certain interval based on how fast the subject(s) are moving in the frame.  For fast moving clouds, cars, or people, I typically set the intervalometer to 2-3 seconds.  For slower moving subjects, I set the intervalometer to 4-6 seconds.  The outcome of my timelapse is typically a smooth clip, with little to no staccato effect.

Now I'm trying to learn how to shoot sunset/sunrise timelapses; the holy grail of all timelapses.  I have some options to achieve this.  I can follow Gunther's instructions from LRTimelapse by manually increasing my ISO or decreasing my shutter speed (for sunset) when the light meter shows that the picture is underexposed.  Conversely, I can decrease the ISO or increase the shutter speed (for sunrise) when the light meter shows that the picture is becoming overexposed.  I can do this manually via the buttons on the camera, which is not preferable since it might cause camera shake; or I can do this wirelessly through DSLRDashboard.  Post-processing for deflickering and stitching the pictures together would be done through LRTimelapse.  By this method, it requires that I buy the full version of LRTimelapse and a wireless router for my camera.  I'm trying to avoid increasing my expenses for this hobby.  As an aside, is my method of creating regular timelapses the correct way or the incorrect way?

Also, someone told me about the 180 degree rule, where the shutter speed is half the interval time.  So for example, if I set my interval to 2 seconds for fast moving clouds, that means (based on this 180 degree method) I have to set my shutter speed to 1 second.  Wouldn't this cause too much motion blur?  And wouldn't this cause a lot of information to be lost between each frame?  Some clarification on this would be awesome.

The second option for me is to use Magic Lantern (ML), and I prefer this option because (apparently) all of the above can be done using the AutoETTR, deflicker, XMP, ramping, etc., modules from ML, not to mention I don't need to spend additional money on software and gadgets to make a holy grail timelapse.  I want to start my studies on holy grail timelapses by learning AutoETTR.  So, the following is my understanding of AutoETTR and its parameter after reading the threads suggested by others.

Trigger Mode: 
I know that I need to set my "Trigger Mode" to "Always on" for timelapses.  From my understanding, this allows the camera to analyze what's happening in the frame and to automatically adjust the ISO and shutter speed accordingly to compensate the light during a sunset or sunrise.  When I enable Live-View, I see the RAW histogram and what is being overexposed.  The RAW histogram has an "E" number and three color-coded circles (i.e., R, G, B circles with numbers inside of them).

Here are my questions for "Trigger Mode" and the "RAW Histogram":
1) Do I need to take test shots in Live-View so that AutoETTR works correctly?  I read somewhere that I have to take 2-3 test shots so that AutoETTR can perform it's analysis.
2) What does the "E" number mean on the RAW histogram? 
3) Is there anything I need to adjust or re-adjust after seeing the "E" number?  If so, what parameters do I need to adjust or re-adjust?
4) I know that the R, G, B circles are the red, green, and blue channels, respectively.  But what does the number inside the circles mean?
5) Is there anything I need to adjust or re-adjust after seeing the number inside the R, G, B, circles?  If so, what parameters do I need to adjust or re-adjust?

Slowest Shutter:
I understand that during a sunset or sunrise timelapse, AutoETTR will change the ISO and shutter speed to compensate for the transition.  In order to ensure that AutoETTR works correctly, I have to set the "Slowest Shutter" to be 2-3 seconds shorter than the interval time. 

Here are my questions for "Slowest Shutter":
1) When the sun is setting, the interval should be fast, maybe around 3-5 seconds?  The interval needs to be longer as it becomes darker, maybe 10 seconds during twilight, and maybe 15-30 seconds to properly expose the stars?  In this respect, is there a way to automatically increase the length of the interval?
2) Also, I assume that the "Slowest Shutter" in AutoETTR would also need to be adjusted, or else AutoETTR won't work.  Is there a way to automatically adjust the "Slowest Shutter"?

Highlight Ignore:
I think this parameter ignores the amount of highlight that is being clipped in the frame.  So, for example, "Highlight Ignore" can be set to a certain percentage to ignore the clipping that occurs in a sunrise or sunset timelapse.

Here are my questions for "Highlight Ignore":
1) "Highlight Ignore" is sent based on percentage.  In sunset timelapses, how do I know how many percentage, or the approximate percentage, to set "Highlight Ignore" so that AutoETTR ignores the clipping in the sun without making everything else dark?
2) Do I adjust the percentage of "Highlight Ignore" based on the RAW historgram?  If so, what do I look at in the RAW histogram to know how many percentage to set "Highlight Ignore"?

That's all the questions I have for now.  I apologize if I sound stupid or simply clueless, but I am so determined to learn how to shoot sunset/sunrise timelapses that not knowing how to is driving me nuts.  I hope someone can answer my questions as if they were explaining it to a 10 year old.  Once I have a good grip on how to shoot a holy grail timelapse using ML, I promise to create a thread that explains everything in the simplest terms possible.

Thank you everyone for your patience.

[dmilligan]

1) Technically no, but the first few frames of your timelapse may jump around in exposure until it settles. If you take a few frames first or simply set the exposure correctly then you can avoid this, hence the suggestion

2) it's the ETTR hint, it tells you how many EVs you need to move the exposure to get a "correct" ETTR exposure. This is the same as AETTR's calculation. When AETTR is 'always on' it will adjust exposure by this much for the next shot.

3) see #2

4) the higher the number the more overexposed pixels there are on that channel.

5) it means you have clipped some channels (overexposed) so you might want to reduce the exposure

Slowest shutter:
One thing that's very hard to do with this type of timelapse is keeping relatively normal or constant shutter angle. You just pretty much have to accept this and keep the interval time constant, say around 30s (there may be some more advanced ways to get around this, but I strong advise you to keep it simple for now to get the hang of it). At the beginning when the sun is still out, this means you're going to have a very tiny shutter angle which theoretically makes for very bad motion blur, but in practice it's really not so bad, esp if you're careful about choosing your subject (for example, landscapes don't have very much motion, so it's not really a big deal). Basically you just can't have your cake and eat it too.

You don't set highlight ignore based on the histogram, you set it based on your scene. For example you can't possibly hope to not clip the sun. It's a 'specular' highlight and you want to ignore it. So a certain number of pixels taken up by the sun (if it's in the frame) need to be ignored. In general the default is pretty good. If you have some type of scene with lots of specular highlights (perhaps a city scene with lots and lots of bright street lights that are 'specular'), you may have to bump up the highlight ignore percent, otherwise just leave it alone.

[tetsusaiga]

Milligan, thank you SO much. That made some sense to me. Wow, that felt good to learn a little more and have some questions answered.

I'm going to read your reply a few more times, and then go back and read those threads again to see if they make more sense to me now.

Do you mind teaching me more after I re-read everything?

Thanks again for the help, it's really appreciated.

[tetsusaiga]

dmilligan:  I just got home and re-read your reply a few times to make sure I really understand what you're saying. 

So the "E" displayed on the RAW histogram in Live-View is a "hint" that basically suggests how much EV I need to add to get a "correct" ETTR exposure.  Is this what you were saying in #2 and #3 of your reply?  Also, I think you were saying that ML's Auto-ETTR will automatically apply that "E" amount displayed on the RAW histogram to the next shot in order to get the "correct" ETTR exposure, right? 

Regarding #4 and #5 of your reply, the number in the circles represents the number of overexposed pixels of a particular channel, right?  You also said that I would need to reduce the exposure to make sure that these pixels are not clipped.  But doesn't AutoETTR do this for me, since it automatically changes the ISO and shutter speed based on the amount of light that is changing in the scene?

About shooting timelapses.  You said that the interval length that I initially set for a timelapse has to be constant throughout the duration of taking the timelapse, right? Can you elaborate on this a little more?  When I first started out shooting timelapses, I read somewhere that a long interval will result in information being lost between each frame, causing a staccato effect.  For example, if there are fast moving clouds in a sunset timelapse, setting the interval to 30 seconds will cause a staccato effect in the clouds. 

For "Highlight Ignore," there is no way we can not clip the sun, since doing so will make everything else dark.  I live in NYC and plan on shooting regular and sunset timelapses.  There are tons of street lights and other bright lights in NYC.  I think these are "specular" highlights right?  In order to make sure that these "specular" highlights are ignored, I need to increase the percentage, or else Auto-ETTR will try to expose these "specular" highlights, which will cause the rest of the scene to be dark.

My goal is to make a timelapse similar to this guy (

)  using only ML.  But I have some questions about his settings.  He said he started with a 1/8000 second exposure? I'm not really sure what he means by a 1/8000 second exposure. Is he talking about his shutter speed, interval length, or something else?  And how did he get the clouds to be so smooth if the length of his interval is long.  Like you said, the interval should be constant, so he must have set a long interval at the beginning of his timelapse in order to capture the stars at night.

Thanks again, dmilligan.  I'll be doing more research and then probably try out what I learned this coming weekend.

[dmilligan]

So the "E" displayed on the RAW histogram...

Correct

Regarding #4 and #5 of your reply, the number in the circles represents the number of overexposed pixels of a particular channel, right?

I don't think it's the number of pixels exactly, but it's proportional to the number of pixels (maybe it's percent? IDK).

You also said that I would need to reduce the exposure to make sure that these pixels are not clipped.  But doesn't AutoETTR do this for me, since it automatically changes the ISO and shutter speed based on the amount of light that is changing in the scene?

Right, so if AETTR is on, then you shouldn't even see this happen b/c AETTR would adjust the exposure down for you. If AETTR is off, and you're 'ETTRing' manually, then this is what you would do.

About shooting timelapses.  You said that the interval length that I initially set for a timelapse has to be constant throughout the duration of taking the timelapse, right? Can you elaborate on this a little more? 

Not really, it has to be constant. I don't know what more I can say about that. It's just the way you have to do it (or at least it's the way you SHOULD do it for now, b/c it's simpler).

When I first started out shooting timelapses, I read somewhere that a long interval will result in information being lost between each frame, causing a staccato effect.  For example, if there are fast moving clouds in a sunset timelapse, setting the interval to 30 seconds will cause a staccato effect in the clouds. 

In theory yes, these type of settings would look "bad". But in practice it's really not so bad, and: *you don't have any choice*

The only way I can really think of overcoming this limitation would be some kind of automated, very strong, variable ND filter. That's way too much complication. Just deal with the 'staccato' it really doesn't look as bad as you think:

There are tons of street lights and other bright lights in NYC.  I think these are "specular" highlights right?  In order to make sure that these "specular" highlights are ignored, I need to increase the percentage, or else Auto-ETTR will try to expose these "specular" highlights, which will cause the rest of the scene to be dark.

Correct

My goal is to make a timelapse similar to this guy (

)  using only ML.  But I have some questions about his settings.  He said he started with a 1/8000 second exposure? I'm not really sure what he means by a 1/8000 second exposure. Is he talking about his shutter speed, interval length, or something else?

If you see a time and a reference to exposure you can be sure the person is talking shutter speed.

And how did he get the clouds to be so smooth if the length of his interval is long. 

Like I said, it's really not as bad as you'd think. Have you actually tried settings like this too see what it looks like? Clouds move very slowly, so with clouds, it's really not a noticeable difference (in my timelapse I posted above, do you see any real difference in cloud motion from day to night? I used a constant interval, so the shutter angle varies drastically from near 0 in the day to near 360 at night). Where you get bad "staccato" effect is with faster moving things like people, birds, cars, etc.

[PaulJBis]

I had replied to tetsusaiga privately, but for the benefit of any newbies, I'll repost here what I wrote to him about what ETTR is, with a bit of expansion:

Basically, and to summarize: digital sensors have more noise in the shadows; therefore, in order to avoid noise, you overexpose your shots as much as possible (without clipping), so that most of your picture is in the brighter side of the histogram (the right). If you prefer the picture darker, you can always darken it later (because you are shooting RAW). That's what ETTR is.

What is AutoETTR? Well, it's a module that calculates automatically what's the highest exposure you can make in any given picture without clipping the highlights. What's considered an "acceptable" amount of clipping? That's what all those parameters and buttons are for, to tune that to your preference.

Why is AutoETTR relevant to day-to-night timelapses? Because in these, the exposure changes with time, and you don't know in advance what the right exposure will be (unless you stay all day and night with a light meter and write down the values for each hour before your shoot ). So you use AutoETTR and let it expose for you.

Will this cause flicker, when AutoETTR changes ISO or shutter? Well, yes, but then you just have to deflicker later using dmilligan's script.

Hope this solves any doubts.

[Audionut]

I don't think it's the number of pixels exactly, but it's proportional to the number of pixels (maybe it's percent? IDK).

The colored circles in the histogram, warn of overexposure in the color channel.  Small circles show a very small amount of exposure, with the circles increasing in size.  When at least 1 % of the pixels of a channel are overexposed, the numbers inside these colored circles, describe the percentage of pixels that have been overexposed in that color channel.

Also, rather then PM'ing a bunch of different people, it would be highly preferable to simply ask questions in this thread.  This solves a number of problems, firstly, you don't duplicate the time from numerous members, and the answers are available in public for everyone to benefit.

[budafilms]

Hi Everybody.
I shoot videos with Mark III MLV RAW, and I will use AETTR.

My worst case scenery is the next: in the street, I ignore specular lights from the street moving the camera out from light spot. I put the T (F) 1.5 (the widest in my lens) I press SET and I get and ISO 6400 and  31/1.  I read in other post about don't get panic with that high ISO in Mark III by A1ex.

Do you recommend shot in that ISO or put a maximum of 3200 - downloading the recommended by ETTR - and overexposed in post?

As you know over ISO 1600 the differences it's huge.

Thanks.

[dmilligan]

Higher ISOs have less noise (per photo-electron). Shooting ISO6400 (which I understand is actually analog on the 5D3, on others it's digital so this advice would be different) is going to be less noisy than shooting ISO3200 or ISO1600 and bumping up to the same level in post.

If you can't get more light to the sensor somehow (wider aperture, slower shutter, add light source to the scene), then it's always going to be best to use the highest (analog) ISO as necessary. In fact this is the whole point of ETTR.

[tetsusaiga]

Sorry about PM-ing a number of people to get help, but I was really eager to find out some answers to my questions.  I figured that some my not see my post, but will see my PM.  I'll post in the thread from now on.  Sorry for the trouble.

dmilligan: Thanks for confirming and answering my questions.  I watched your timelapses and they all look really good.  So I'm guessing in each of the day-to-night timelapses you had a constant interval throughout the whole time?  What was your interval and shutter speed for the Fall Creek timelapse?  I'm going to try my best to find some time tomorrow and try out a long interval time and shutter speed to see how the timelapse comes out.  Do you have a recommendation for setting the shutter speed and interval for slow, medium, and fast moving clouds?  If not, I can wing it and see how the timelapse comes out.

PaulJBis:  Thanks for answering my questions.  While we were talking about staccato, I was shocked that you said your interval was anywhere from 15 seconds to 30 seconds throughout a single timelapse.  The reason I was shocked is because I kind of pieced a lot of information from my research and it said to use faster intervals for medium to fast moving subjects.  For instance, this timelapse tutorial (

at 3:41 to 4:45 makes the distinction between smooth and jerky timelapses.  He states, essentially, that fast moving subjects require a shorter interval to obtain a smooth timelapse, while a longer interval will result in a more jerky/choppy timelapse.  This is what I was talking about in my cloud example in our PM and, having made smooth timelapses in this way, I was shocked to hear that a smooth timelapse can also be achieved through a long interval.  I'm going to shoot a timelapse, any suggestions for slow, medium, and fast moving clouds?

By the way, I'm currently using the trial version of LRTimelapse to make my regular timelapses.  Do you guys suggest buying the full version of LRTimelapse?  I was thinking of purchasing it because I'm capped at 400 frames and I'm not sure that's enough for a day-to-night timelapse.  On the flip side, do you guys suggest I subscribe to Adobe CC instead?

Thanks again for everyone's help and patience.

[PaulJBis]

PaulJBis:  Thanks for answering my questions.  While we were talking about staccato, I was shocked that you said your interval was anywhere from 15 seconds to 30 seconds throughout a single timelapse.

No, that's not what I said. I said that I normally use intervals of 30 sec. or 1 min., but what I do NOT do is to change interval times within a single timelapse. If I start with a 30 sec. interval, I stick to it until the end.

Edit: or, in other words, what dmilligan said.

[dmilligan]

What was your interval and shutter speed for the Fall Creek timelapse?

details in this post: http://www.magiclantern.fm/forum/index.php?topic=12330.0
(I'm pretty sure the shutter speed varied from 1/8000 all the way to 32", so the entire range of what the camera is capable of)

Do you have a recommendation for setting the shutter speed and interval for slow, medium, and fast moving clouds?

I wouldn't really say clouds move slow, medium or fast, I would say they move slow, very slow, or very very slow. That's why I would say you will almost never have an issue with this, clouds just move too slowly for it to matter.

You seem to me to be overly concerned with setting your interval to get "good/smooth motion". For me, I consider this a secondary goal that is subservient to other more important things. For example, it's called a time-lapse for a reason, time is lapsing. In general the more time that lapses, the more interesting the time-lapse is. I would much rather see an interesting, fast moving time-lapse that is "choppy", than a slow boring one that is "smooth".

I will always set my interval first without regard to shutter speed or motion, based on how much time I want to capture or feel that the scene calls for being captured. Then after I decide that, I can try to get "good" motion with a shutter speed that gives me reasonable shutter angle (the goal being 180, i.e. interval time = 2 * shutter speed). Maybe I can achieve that by adjusting aperture or using an ND filter. But if I can't then I just let it be. My interval that I select does not serve the purpose of "good motion", instead the shutter speed serves the interval I've chosen and "good motion". And if I can't get the shutter speed there, well then I don't worry about it.

In general I never use any interval time less than around 15 to 30s when timelapsing nature or landscapes (any slower and I just feel the result is too boring). The only thing that might call for a shorter interval is something that changes more rapidly, such as people moving about, cars, boats, vehicles, building or assembling things (like a guy building a lego set, etc.), or perhaps a fast moving storm.

It helps to have a good strong ND filter.

Of course all of this is just my subjective opinion, so take what you will and disregard the rest.

[tetsusaiga]

PaulJBis:  Sorry, I should have phrased what I said a little differently.  What I meant to say is that I was shocked your interval was either 15 seconds OR 30 seconds for a single timelapse.  That was my mistake for phrasing it incorrectly.  But I now understand that a single timelapse can have a long interval, as opposed to my initial misconception of setting a short interval to make sure I capture enough information.  Thanks for clearing that up for me. 

dmilligan:  You're right that clouds do move slow, very slow, or very very slow  .  To answer your implied question, yes, I am overly concerned with getting "smooth" timelapses.  I think my concern stems from the tutorial (https://www.youtube.com/watch?v=W4SzZXLiyvk) at 3:40 to 4:30.  Maybe I interpreted this section of the tutorial incorrectly, or applied it to the wrong scenarios.

Your approach to what you set first in your timelapse is opposite of mine, which I now know is the incorrect way.  I'll try to analyze what I'm shooting and then set my interval first, followed by the shutter speed and ISO.  I like taking timelapses of nature or landscape, so I'll remember to set my interval anywhere between 15 seconds to 30 seconds.  I'm going to be camping in Pennsylvania, upstate New York, Lake George, or Delaware next month, so I kind of want to have this down before I go.

With respect to the 180 rule, I just want to make I sure understand it.  For example, if my interval is set to 30 seconds, I should divide the interval by 2 to get my approximate shutter speed, right?  In this example, an approximate shutter speed of 15 seconds.    But this 180 rule shouldn't matter anymore once I learn how to use AutoETTR, right?, since AutoETTR automatically adjusts the shutter speed and ISO based on the available light in the scene.

My last question for the day is about your comment on the shutter speed varying from 1/8000 all the way to 32 seconds.  I'm assuming that it was AutoETTR that varied your shutter speed, right?

Since everything I pieced together in the past year about timelapse has essentially went out the window, I'm going to try and shoot a regular daytime timelapse with everyone's' comments in mind.  Hopefully the final product will come out to what I expect.  I eventually want to learn how to shoot sunset/sunrise timelapses, and hopefully I will soon.  Dmilligan, your timelapse link above is really nice, especially at the 30 second mark where the cloud appears out of now where and moves across the sky.

I hope that you guys can continue to educate me and correct me where I'm wrong, like you did above.  Thanks again and it's really appreciated.