I've been shooting a timelapse & hyperlapse project the past few weeks, and I've noticed something strange:
When processing the photos in lightroom, even though the ISO, Shutter & aperture are exactly the same, lightroom interprets the images differently, and applies some kind of smart contrast to the images when imported, to probably even out everything nicely and give you a better result. This is a known "issue" (it is not an issue unless for timelapses).
This happens even BEFORE I have done any edits, and is NOT wanted during a timelapse as it causes flickers. In this given example, only edits I did was -20 highlights and +20 shadows, but as I said it even occers with 0 edits: https://www.youtube.com/watch?v=4eYJZgarKz0
The images on my camera however, and converted to jpeg, do NOT have this flicker. The flicker only seems to appear when something comes into frame.
So, here's my question: is there any way to disable it?
Please note, it is absolutely not aperture flicker or even any sort of flicker caused while shooting. It is caused in the post production or in the interpretation of the files.
Copy link to clipboard
Lightroom does some automatic highlight recovery when rendering raw files.
The car's headlights are probably blown, causing Lightroom to try to recover some detail.
This will also affect other highlights in the image, like the sky.
It the other video, I think it's the changing sky that causes it.
There is no way to turn this off, so you'll have to edit the individual images to correct for it.
Alternatively, you can try a different raw converter.
Thank you for your response. Lightroom indeed seems to do automatic highlight recovery as well as automatic shadow recovery.
I partially disagree with your comment though: what is causing it is not the sky, but the pedestrians walking by in the second video. Strange, no?
This does NOT happen incamera, or processing the raw to jpeg in camera, or using capture one. It only happens in Lightroom, which is still my raw converter of choice. It somehow flattens (aka brightens) the whole image.
Is there a workaround?
If not, are there lightroom versions which do not have this?
Copy link to clipboard
Did you use some sort of automatic setting on the camera such as aperture priority or shutter priority? I wonder if this anomaly could be minimized by setting the camera to manual exposure. Don't know for sure, just wondering.
No, of course not. Aperture was locked (screwed off lens) and shot in live view mode.
Please believe me when I say it is Adobe Light room pr at least post processing related. Everything was shot as manual as can be.
Copy link to clipboard
Although you mentioned it was before applying any edits, because the Lightroom Classic defaults are customizable are the default import settings applying any of the image-adaptive adjustments such as Highlights, Shadows, Texture, Clarity, and Dehaze, or are all of those set to zero for all images?
I indeed applied shadows and highlights but thats pretty much it. No texture or dehaze or anything else. But even without it, the histogram I see before apllying anything is seemingly different from one shot to another in some cases, and this is NOT the case when processen in capture one or when i convert images in camera.
As you can see, it only happens when something big enters the foreground, i.e. the car ot the pedestrians.
Out of curiosity, what were the values used for Shadows and Highlights? I’m interested in how much it took for the frame-to-frame changes to be so obvious.
All of the adjustments I noted must be used sparingly for images in a time lapse sequence. I use the LRTimelapse application, and the developer specifically warns about this in one of his blog posts. Although it’s primarily about Dehaze, he also mentions Shadows and Highlights in the same post.
From the link:
Each image in a time lapse sequence is a bit different from the last. Depending on the changes between images, the Dehaze filter might work stronger or weaker on subsequent images and even deliver different looks with the same settings. Small changes between images might be sufficient to introduce a large change after applying the filter. This means a rather smooth sequence might get some serious flicker introduced when you apply the Dehaze filter. The same applies to Clarity and to a certain extent to Blacks/Whites and even Shadows/Highlights, if you use them to great extent. The more intelligent and context sensitive the tools in Lightroom are, the stronger you use them, the likelier you will get nasty contrast flicker introduced by those tools.
This is not necessarily a bug; Lightroom Classic is designed as a still image editor and I don’t think it claims to have any built-in support for time lapse sequences. It is possible to submit a feature request for some kind of option that would let you define and account for image sequences when applying the image-adaptive adjustments.
One answer for preventing those contrast changes is to avoid using the features that cause them. For example, instead of using the image-adaptive Shadows and Highlights adjustments, adjust shadows and highlights as much as you can with the Tone Curve. The Tone Curve is not image-adaptive, so it should apply the same adjustment to every image without causing flicker.
Sometimes I do want to apply Shadows and Highlights adjustments to a time lapse, but if the amount I want to apply would risk flicker, I render it out from Lightroom Classic without them, and then I apply them later while editing further in Adobe Premiere Pro — because its Shadows and Highlights adjustments are video-aware.
First and foremost: I absolutely agree with you Lightroom is a still frame editor and if used for timelapse, it should be used with care.
BUT: please please please, for the love of god, understand, there is something funky going on here.
The contrast is applied before the pictures are even processable. There is some kind of "auto" highlight recovery and shadow recovery that is based on the picture itself and nothing else. This causes a flatter histogram as well.
Check this out, no adjustments, imported straight from the SD card: https://www.youtube.com/watch?v=AUMSgHXOc5A
This is NOT the case when processed in Capture One or incamera.
Yes that is by design. In the last few process versions there is an automatic recovery of blown highlights. This will impact only the extreme highlights. The other change you see is simply the application of the Adobe color profile to your raw data. That will render differently than the built-in jpeg engine and will impact the entire image. If you are on a Canikony camera, you can use the camera standard profile to minimize that shift.
I have also tried different camera profiles: adobe standard, adobe vivid, camera standard, camera flat, camera vivid and what not.
I literally tried every setting in adobe, yet i have not found a solution.
I absolutely love the versatility of adobe lightroom, but I can't seem to image i'm the only one running into this issue? In one of the other videos you can find above, there is obviously some kind of exposure compensation or contrast compensation going on, which I would love to disable.
Actually, that video doesn’t indicate that there are no edits, because it shows the Quick Develop panel in the Library module which doesn’t display the current values for us to see. Can you please upload a video or screen shot showing the Basic panel in the Develop module for adjacent images?
Sorry, that was my fault…I should have watched more of the video. I now see that it switched to the Develop module where there are no edits. Thanks for your patience as we requested more information, we just wanted to be sure. (The internet has become a place where people believe claims with insufficient evidence, so many of us prefer to err on the side of diligence.)
If I stare at the grass or the trees, it’s difficult to see any difference. I do see a little change in the sky, not sure if the clouds could have changed enough during the time lapse interval.
Now that we cleared up more of that, I don’t have a quick answer for this, nor for why you don’t see it in Capture One. You already mentioned ruling out aperture flicker, and if that means the aperture was held open the entire time (not just manual aperture) then I’ll accept that. I’ll have to think about it some more; for example, I don’t think the selected camera profile should affect this.
Copy link to clipboard
This happens because you applied highlights and shadows. They can cause flickering in timelapse sequences. This happens because highlights and shadows (as well as blacks and whites) are HDR techniques that dynamically find those regions in an image that it should apply to based on an analysis of the entire image. if you get a bright object in the image for example or have some noise around, the whole image will get impacted by those sliders. Only solution is to not use those sliders.
Yes, many of the LrC / ACR image adjustments are adaptive to the 'accidental' content of each image.
One approach is to forego using nearly all these Basic panel adjustments, when constant processing is necessary across many frames of a sequence. But if the point of the exercise is to achieve desired results, this is not very productive IMO! And relying on e.g. Tone Curve to do the job alone, would I think be a poor strategy.
So another approach turns to different forms of 'legacy' LR / ACR processing instead, by setting a much earlier Process Version for all the images in this sequence. AFAIK the pre-'2012' adjustments are not image-adaptive: so, a constant setting should not alter how it affects a given input value, according to what else may have appeared or disappeared in a particular picture, nearby to that.
I have also looked at the pre 2012 exposure adjustments. But it is no longer possible to change them by 'year', only by version, ranging from 1 to 5, but even then, the contrast shift still occurs..
I understand and agree to use the sliders with caution, but how do you explain this:
No edits, nothing, is applied. And still, different gamma: look at the shadows in the trees.
If this is user error: why is this not the case in Capture One?
I've watched this video many times and I don't really see what you are referring to. I see changes between the two images but nothing in the bulk of the image, just the car appearing and disappearing. the shadows in the trees might be subtly different but hard to tell. IS that what you are referring to? It is very subtle and might be just noise from wind blowing leaves around. Do you actually see this difference in an exported jpeg? The zoomed out view in Lightroom is not to be trusted when there is a lot of small detail as it completely depends on how large the preview image is that was created because the image needs to be scaled down to show. If the difference survives jpeg export to the desired size, I would try to turn the process version down to process 2 or 3, which should not have any highlight or shadow recovery built-in.
It's a subtle difference, but it's there.
I have decreased the process version down to 2 or 3 and it still occurs. As mentioned many times, it happens even without any edits whatsoever, and that's what I'm showing you in the video. It's small, but it's there.
The car appearing is what I believe that causes the "lesser" contrast, as the car is black, so the images' contrast must be decreased which causes an overall gamma change.
I am a professional photographer and videographer for 8 years, I own Nikon, Fuji & sony cameras and have used lightroom since as far as I can remember. I am not judging the small image previews. I am judging the full exported jpg sequence with zero edits that has flicker yet has the same settings and aperture locked.
Is there some kind of witchcraft going on? Again: I do the exact same processing in Capture One, and there's no flickers. I do the same processing in Lightroom: flickers.
What kind of lens do you use? My experience is that it is virtually impossible to avoid any flicker if it is not one of the old fashioned ones where you lock the aperture in place or when you shoot wide open every time as the aperture on modern lenses doesn't hit the set point exactly every time. You will notice even a few percent difference in exact aperture. You're taking these images at f/11 so you could definitely be dealing with this issue.
As far as I know there is no built-in mechanism in Lightroom's code that could do any automatic contrast control if all your sliders are at zero. It just pumps the raw data through the profile and in the most recent process versions does a bit of highlight recovery which only affects the top highlights but should never affect the shadows. I've never seen this behavior in my timelapses but I don't do an awful lot of them.
It's the 70-200, but it was locked. I don't know why I have to repeat this so many times? I know exactly this is not user related. Why?
Because of 3 reasons:
1) the issue does not show up in camera or in Capture One
2) the aperture was locked so there was no aperture shift possible
3) the histogram clearly indicates a contrast squish and a contrast stretch. Aperture flicker would give us an exposure shift left or right.
Histograms in lightroom are based on the current crop. Check out this screenshot, cropped only to the sky and trees. And still there is contrast squish or stretch.
Sorry I missed that. I see now that you screwed off the lens, which should completely get rid that issue. I seen the shift in your video. No clue how that could happen - Lightroom/camera raw should not be doing any of this. Any chance you can share the two raw images? Do you have any dynamic settings enabled in camera such as active D-lighting?
Looked a bit deeper and I see the same effect in some images taken fractions of a second after another in burst mode with every develop setting zeroed. The histogram jumps around. Nothing like that when using Nikon's software or the built in raw converter in Mac OS X. I think you are right that there is some degree of automatic compensation of contrast happening. I never noticed this before as it is very subtle. You would notice this in a timelapse for sure.
Edit: I was wrong. The two images had diferent profiles applied. They show no shift when their develop settings are completely identical.