Trial and error with time lapse video 1


I’d like to share some of my experiences in learning how to do time-lapse properly, above all using HDR photography. I hope you can find this information useful and can save you some time before trying to do stuff that simply doesn’t work from the beginning.

I’ll also use this post to review the apps, software and equipment I’ve used in order to achieve good results.

 

By good results I mean the following video:

But be aware! that video is still not perfect! so let’s start first from the beginning and at the very end I’ll reveal what else is wrong with this video so you and me can try to improve our time-lapse technique.

Speeding up videos isn’t the way to go

Time-lapse video is putting together lots of pictures taken at specific intervals (i.e. one frame every 0.5 seconds) and put together in a way that they play at the frame rate typical of a normal video (25fps in this post as we are writing from Europe). The final result is a high speed version of the reality that was photographed/filmed… then, why all the hassle and not making a very long video and speed it up? After all, this would be the result:

If we record that scene during 4 hours and we make the video 80 times faster we get what we are looking for. In fact, we have a pretty flexible raw material if we wish to change the speed.

That video was filmed with the following settings:

  • 1920×1080; being this the maximum frame size we have so there is no room for cropping.
  • 50 fps. This factor is pretty good, plenty of frames to do whatever we want with the time. From speeding it up as above to slowing it down resulting in a very smooth half speed slow motion if the final video is at 25 fps.
  • H.264 video compression. Here is where the problem is. Even if we have lots of frames, each of them is highly compressed, not even each of them as MPG based video codecs compress sequences of frames to save in space! Modifying the image in post-production is subject to important quality loss if we go too far into adjustments. Nevertheless, the video above has a heavy colour treatment… so it is still doable. More image treatment would mean more quality loss. In short, a frame from a video is a low quality start if we compare it with a whole photograph with all its pixels (not to mention if we work with RAW pictures), its size and its low compression.

Some people will add that by taking photographs instead of video you will consume less memory. That’s not entirely true if getting the maximum quality out of each frame is your aim. What will happen is that instead of having 42GB filled with a 3:37 hours video at 50 fps (highly compressed), by taking photographs in a time-lapse you will have 42GB with less frames at much more quality.

The video compression depends on the camcorder and format you use. For the video above I used a Panasonic HDC-TM900. Obviously, higher end cameras will invest in more card memory in order to keep images at higher quality. But overall, the following numbers won’t vary dramatically amongst camcorders as we are comparing still pictures with video frames and technology still needs some time until both are indistinguishable (which is something unlikely to happen any time soon).

So let’s have a look at the following table that compares amount of frames (still images) between a camcorder and photographic cameras in order to get the same final video length:

 Source frames (or stills) taken Memory consumed Post-production speed up  Final video length mins/frames
Panasonic HDC-TM900
(Video camcorder recording at 25 fps with H.264 codec)
326150 frames
(3:37:26 hours)
42 GB
(0.13 MB/fr)
x80 00:05:45
8625 frames
Canon 7D
JPG* stills at maximum quality
 8625 35.4 GB
(4.20 MB/fr)
x1 00:05:45
8625 frames
Canon 7D
RAW* files
 8625 202 GB
(24 MB/fr)
x1  00:05:45
8625 frames
GoPro Hero 3
JPG* time-lapse
 8625 27 GB
(3.20 MB/fr)
x1  00:05:45
8625 frames
*JPG and RAW size depends on the complexity of the picture. The more complex (colours, shapes, etc.) the more size consumed. For this example, the differences between video and photographs is so big that the JPG/RAW size wouldn't affect the purpose of this example.

So as you can see, if given the option, we will want to invest in quality using a Canon 7D taking the RAW pictures we need for the final length of the video instead of wasting so many frames recording a really long video at very high compression. Even though, the memory needed to get the same final length with RAW pictures is much higher (202 GB) than the memory needed to capture an almost 4 hours video (42 GB)!

In short: Quality is the reason we choose timelapse instead of video.

Pixel motion may be cool sometimes (and a short mention to the important frame blending)

You will observe in many timelapse videos that movement of elements such as people, tree leaves, and water waves are rather clumsy. They jump from one position to another with no transition in between. This happens in videos that contain an amount of activity that the frame rate (timelapse interval) can’t capture.

This is not a real problem. It is up to the author of the video to decide if leaving the timelapse as it is or applying some post-production effects that smooth the action.

Let’s compare the outputs. This timelapse video was made with a GoPro camera and it doesn’t have any post-production effect to smooth the action:

You can add a frame blending effect using Adobe After Effects, or you can go beyond that and apply pixel motion. This is how the video above looks using pixel motion with After Effects:

And this is a short tutorial on how to combine pixel motion and motion blur in After Effects in order to get the results shown above: 

The most effective way to smooth the video is using After Effects Frame Blending. As shown in the video tutorial above, try to export the video with a frame blending (no need to add Motion Blur) and observe if the results are better. As always, there isn’t a rule when it comes to your desired aims.

Frame Blending in After Effects is a good option to smooth sea waves and tree branch leaves shaking

Frame Blending in After Effects is a good option to smooth sea waves and tree branch leaves shaking

 

The following is a high speed video with pixel morion effect. I personally like the effect on this one as there isn’t much movement of people but when it happens, pixel motion combined with motion blur creates that “sucking” effect that sometimes is fun to watch:

Avoid automatic settings

We all know it is wrong to use automatic settings! The following timelapse is a good example of what to avoid for a number of reasons:

Mistake 1: Using a GoPro for a sunset

GoPro cameras are excellent for timelapse. But they aren’t when the main subject is a sunset and nothing around it is attractive. The scenery is nice, but nothing really important is happening around the sun to keep the viewer’s attention. The problem isn’t the GoPro, the problem is the wide lens. A sunset is better filmed with a telephoto lens if it is the sun only what we want to show. Now, if we have a nice set of clouds and a more complex scenery, then a wider lens will work because we will show a beautiful timelapse of the sun interacting with its surroundings.

Mistake 2: The ocean                          

If timelapse is appealing, it is because it brings us a representation of the reality that we don’t see in real life: Playing with the flow of time. Observing the movement of the clouds; shadows covering mountains; the movement of the sun, the moon, planets and stars… Seeing all that at high speed is revealing and relaxing. But seeing waves moving so fast in such a clumsy way is not (try to apply Frame Blending in post-production and it may look better). Deciding what we want to show is crucial. Let’s try to bring the image further away from the waves so the ocean can show other aspects we usually don’t see. Have you observed how our first video reveals the interaction of the wind with the ocean? I discovered that after assembling the timelapse. Here it is:

Mistake 3: Automatic exposure

GoPro cameras don’t allow manual exposure control. So this is what we have. In any case, regardless of the camera, attempting to create a timelapse of a sunset with automatic exposure is pointless. The sunset above takes so long to get dark because the camera was constantly modifying shutter speed, aperture and gain in order to keep the scene well lit. That is, well lit according to the camera. What we wanted was to see the whole scene become darker gradually and naturally as our eyes would see it.

Mistake 4: Automatic white balance

Although GoPro allows manual white balance, I made the mistake of leaving it automatic. Again, let’s don’t use automatic white balance in any timelapse and less when we deal with sunsets. As you can see, the sky colour changes several times. It becomes red, but the camera corrects it and it becomes white again, to become purple and then the camera tries to correct it again. In short, the timelapse is not showing the nice colour change characteristic of a sunset.

Sunsets are probably the hardest scenes to capture in timelapse and we need a range of equipment and techniques to control the light. A lot more on this later!

 

Set manual focus on your camera!

If you want to ruin a timelapse, use autofocus. GoPro and mobile phones won’t let you manual focus, but when dealing with other cameras, stick to manual. Otherwise your camera will try to focus and refocus each time it takes a photo. It will take longer to take the picture; the focus can point to different places and the framing will change. Most probably, after making this mistake, you’ll find that your video is unusable. Like this one:

Timelapse videos are better when they have a reason to exist

Making a timelapse is fun. But creating one doesn’t assure that it will be attractive. Timelapse is like any other video technique. It is all about the message you want to send. Timelapse is the container of the story.

One of the ideas I came up when I got my GoPro and I was digging on Amazon looking for gadgets and extensions for the camera was the now named “Around the table” project. Here is the series:

In order to make this I needed:

GoPro tripod mount

GoPro tripod mount

GoPro Hero 3 Black edition

GoPro Hero 3 Black edition

Veho VCC-100-XL MUVI X-Lapse 360 Degree Photography and Timelapse Accessory

Veho VCC-100-XL MUVI X-Lapse 360 Degree Photography and Timelapse Accessory

Episode 2 of Around the Table contains more variety of shots. Although 360 degrees timelapses around the tables in coffee shops are still the main visual element, the video is using timelapse as a technique to deliver the message rather than being the message itself. This second part is more enjoyable in my opinion:

Timelapse videos are 4K resolution videos at no extra cost      

The resulting video from a timelapse can be a 4K video. This is because we are taking photos that happen to be as big as 4K (or UltraHD) video frames. The first part of Around the Table is a 4K video:

The photo size of the GoPro Hero 3 Black edition is 4000×3000 ppi (pixels per inch) at 72 ppi resolution (72 ppi is video resolution. Don’t bang your head with Apple or Windows different resolutions or anything like that. Just think in 72 ppi and your life will be easier). A Full HD video is composed of frames sized 1920×1080; and Ultra High Definition or 4K video is composed of frames sized 4096×2160. Ultra HD is the term used for consumer 4K displays and they show resolutions of 3840×2160. In this post we are dealing with Ultra High Definition at 3840×2160.

So, the resolution of the GoPro photo fits a 4K video. We end up with a frame that is using nearly the full with of the original sequence of photographs and still leaves some room to recompose the photo vertically as we have 840 vertical pixels that we won’t use (3000 – 2160 = 840).

Here I spent some time comparing frame sizes and how the original timelapse photos can be used for 4K, 1080 and 720 final videos:

HD comparison

So we have plenty of freedom. We can even rotate and scale the picture until certain degree to achieve the perfect composition.

 

RAW vs. JPG

By default, most consumer cameras will take photographs in JPG format. DSLRs and prosumer cameras will let you take photos in RAW format. RAW is an uncompressed picture that contains all possible data regarding colour, luminosity, metadata, etc. The size of just one RAW picture can be 25MB for a 18MP camera such as Canon 7D. The more megapixels (MP) and the more detail, the bigger the size of the RAW file.

Since JPG is a format that compresses the picture provoking quality loss, there isn’t much to say really: RAW is the way to go. The problem is that as we saw in the table above, a JPG picture can take about 3.5MB while its RAW equivalent will take about 7 times more disk space. That translated into price means that RAW is more expensive. A 128GB CompactFlash card can easily reach £200. And believe me, when we start dealing with HDR timelapse, you will need that amount of space.

But let’s say we have a 16GB CompactFlash card in a Canon 7D 18MP DSLR. The camera is set to take a RAW picture as well as a fine quality JPG still for each photo. It seems the card will be able to store 448 pictures. That’s almost 18 seconds of final video. If our timelapse is made of one photo every 2 seconds, we can plan a timelapse session of 15 minutes.

What happens in 15 minutes that is worth taking a timelapse of? Let’s capture the movement of the clouds in a windy day!

 

Calculating final length, frames and working sessions

Before going to the battleground with our equipment, it is usually good to plan how long we will need to make the video as that factor will be linked to the final length of the video and the amount of disk memory we will use. There is no formula that tells us what interval the timelapse should be set to. This factor depends on what we want to capture and it is the most crucial one, as it will inform about the rest of factors.

The following are different intervals for different case scenarios. I write them as they come to my mind from previous experiences. But I may modify them or add more when I gain more experience in this area.

 Scene Interval Final length NO. of pictures
(25 fps)
Time-lapse session Storage RAW (25MB) Storage JPG (3MB)
Sunset  10 secs  20 secs  500  83 mins.  12.2 GB  1.5 GB
Clouds moving at a slow pace  5 secs  20 secs  500  42 mins.  12.2 GB  1.5 GB
Clouds moving at a fast pace  2 secs  20 secs  500  16 mins.  12.2 GB  1.5 GB
People in the city  1 sec  20 secs  500  8 mins  12.2 GB  1.5 GB
Formula:  Common sense and experience  Your guess  25 fps x Final length in secs.  (NO. of pictures x Interval in secs.) / (60 secs.)  (MB per photo x NO. of pictures) / (1024MB)  (MB per photo x NO. of pictures) / (1024MB)

If you are like me, you will probably start without any calculations. That isn’t entirely bad because you’ll find yourself in trouble and will have to think hard why all went wrong. A good example of disaster is the following:

Firstly, that shouldn’t have been a timelapse. The clouds where moving so fast that the timelapse was completely pointless. Secondly, in order to capture the detail of the moon’s surface together with the details of the clouds, I needed to do a HDR timelapse taking 3 different photographs of the same moment with different exposures. The clouds where moving so fast that each photo from the HDR had a different set of clouds. To make things worse, at least one of the HDR photos had to be long exposure in order to get the detail of the clouds in the dark, so the interval between photos was even higher making the timelapse and the HDR totally disastrous. Here you go, not everything is success in life 🙂

                                                                

What do we need to create a timelapse video with a DSLR

The essentials are the following:

  • A DSLR camera. I’ll be using a Canon 7D.
  • The camera lens. Important decision if you are serious about what you want to film. For this example, the Canon will have a Sigma 10-20mm f 1:3.5.
  • A stable tripod. I will use an old version of GorillaPod Focus. This isn’t the best example of stable tripod but I’m pretty sure it will work fine.
  • A timer remote controller for your DSLR. I will be using an Aputure Timer Remote Controller AP-TR3C that costs £20, a lot less than the official Canon and it does a great job. There are lots of compatible timer remote controllers out there and all of them look similar. That makes me think that there is one manufacturer and many companies that just stamp their logo on them.
  • Adobe Bridge and Adobe Photoshop for RAW batch editing, so we can fine tune all our pictures at once.
  • Video editing software that can import photographic sequences and export them as a video file. I will use Adobe Premiere CC and GoPro Studio.

The not so essential elements:

So once we have all this stuff ready, we just do the following:

And this is the final result in 4K ultraHD:

Adding camera movement with a motorised slider: the dolly movement

If we add a dolly movement to our timelapse video, it will go a few levels up! You can even combine the dolly movement with a panning. It just increases the value of your production. Making a timelapse with dolly movement is currently relatively inexpensive. You will still spend over a grand in a slider.

I have used a Varavon slider which seemed to have a market entry price of £600. It has gone up to £1500. To be honest, I ended up very disappointed with the Varavon slider. The Timeroid Body, the machine that makes the camera slide, got broken and Varavon technical support is completely unresponsive. So even if the cost of this set was “just” £600, that was  lost money that could have been gone towards purchasing a good quality slider. On top of all, the item that got broken has been discontinued so there is no way to replace it. Really bad service from both Varavon and their UK distributors CameraKings, wich took 6 months to respond. Watch where you buy media equipment as warranty and quality can be a good investment and being cheap just brings problems on the long run.

In short, although it is true that there are more and more professional accessories becoming available for amateurs, do a very good research before spending your money in bad quality equipment. Instead you could just have saved that money and waited for a better product.

time-lapse-slider

The timelapse slider will have at least three parts:

  • The slider. The longer the better.
  • The motor. However the configuration, it will be attached to the camera and the slider. It moves the camera along the slider by rotating an engine plugged with strings to both sides of the slider.
  • The remote control for the body and the camera that will synchronize the movement

You will need to do some extra calculations when setting up your timelapse plan. The remote control of the slider will move the body during the period the camera is not taking photos. The remote control will know when to move, but you will have to set an amount of time for the body to stay still, enough to cover the exposure time of the camera and the number of pictures the camera will take (in case of HDR).

Here is an example of timelapse with dolly movement taken with a slider AND using HDR (High Dynamic Range) photography. We will shortly analyse this video in order to learn how to use a slider and make HDR timelapse, probably the coolest timelapse technique.

The timelapse above is far from being optimal. Before praising it, the following are the errors made when making it:

  • Photographs were taken in JPG format. Ideally they should be RAW but sometimes factors such as time and memory available restrict desirable techniques.
  • Three cuts were done during the timelapse recording in order to adjust the camera’s exposure time. I fixed the aperture and shutter speed (manual) but during the shooting I realised that the camera was getting short with light when the sun was gone. Since this was a HDR timelapse using the bracketing option of my Canon 7D (more about this in the following section “HDR timelapse”), I should have set the camera with aperture priority (Av). Additional problems come with Av though! I’ll explore this next time I make a sunset HDR timelapse.

Even with the errors listed above, the timelapse accomplishes its function and I haven’t met anybody able to spot the use of JPG format (practically impossible) or point at the three cuts which are more noticeable if you pay more attention.

Now, how to setup the slider! Easy:

  1. Determine total exposure time. At the beginning of the timelapse, three photos were taken (bracketing photography): shutter speeds of 1/20, 1/3 and 3 seconds. Rounding up and adding some safety to our time range, that’s 5 seconds.
  2. Determine your desired interval time between photos. Checking the table above, 10 seconds seems like a good interval for a sunset.
  3. Calculate your final interval, which is the time the motor of the slider will stay still. Since 5 seconds is the minimum according to our exposure time, that fits perfectly in the 10 seconds interval recommended for sunsets.
  4. Assemble your gear, camera and everything else.
  5. Make sure you have enough batteries for the slider’s motor and your camera.
  6. Connect the slider’s remote control to the camera and the slider and…
  7. Input the timelapse settings in the remote control:
    • Final interval, as calculated above, in this case it will be 10 seconds.
    • Slider length, will determine when the motor will stop.
    • What length in millimetres the motor will move the camera between intervals.
  8. There might be more settings, and the way the remote control works depends so much in the slider you bought. But above are the very basic settings that will act asvariables and will provide you with further useful information:
    • Number of photographs that the camera will take (in the video above that number is multiplied by three due to the 3 photos bracketing photography technique).
    • Time the whole shooting session will last.

For the timelapse above, I interrupted the shooting two times in order to change exposure settings in the camera. After sunset, the total exposure time reached 17 seconds. This explains the change of speed of the subjects in the video.

HDR timelapse

HDR photography is the short form of High Dynamic Range photography. It is initially aimed to create one photograph with even exposure in the whole scene. It is useful when taking photographs of sunsets or other scenes where there are big differences of light intensity between foreground and background. So if you see a nice scene to take a photo of and you want to show everything evenly exposed, HDR is your friend.

There are point-and-shot cameras and mobile apps that can do HDR photography with immediate results. But like always, the best results are achieved with a good DSLR using RAW format. The processing with a DSLR needs to be done afterwards using specialised software for this purpose.

Straight to the point, this is what we do:

Set your camera for bracketing photography. Each camera will have this option in a different place in the menu. A canon 7D will look like this. You will also have to set your camera’s timer to 2 seconds since this is the way a Canon 7D takes the three photos.

 Canon-7D-Menu2  Canon-7D-Bracketing

The centre of the bracketing represents the average exposure, and then you can select how many steps the camera should underexpose and overexpose in the other two photos. You’ll have to test this according to the scene you want to capture.

We press the shutter, wait 2 seconds for the camera timer and the camera will take 3 consecutive photos with the three different exposures.

 HDR1  HDR2  HDR3
 ISO 100, f 22, 1/20   ISO 100, f 22, 0.3   ISO 100, f 22, 3

Using a HDR processing software we merge the photos in one and create this:

HDRfinal

 

Since HDR became popular there is a current trend of haters. The reason is that HDR has been “overdone” by some photographers to the point that some people associate HDR to unrealistic, cartoonish, paint-like, surreal photography.

HDR is just a technique and it is up to the photographer how to use it. And the image above demonstrates that a HDR photo can be perfectly realistic. In fact, probably nobody could tell that the resulting photo from the example above is the result of HDR.

In my experience, software like Lightroom and Photoshop produce those unrealistic results (or at least I don’t know how to use them properly for HDR). The software I personally like is Photomatix Pro. It is very flexible with quick settings as well as very intricate manual controls.

The way this is applied to timelapse is by taking bracketing photographs for each frame. So take this into account when planning the amount of memory you’ll need.

If your camera requires timer for taking bracketing photography, add that time to the interval time between shots.

You also have the option to take more than 3 photos for each scene. High range cameras such as the Canon 5D gives you the option of up to 7 photos! But if you don’t have such a DSLR you can connect your computer to the camera and use software to control it or, connect your smartphone with an app for the purpose. I personally like DSLR Controller for timelapse, bracketing photography and much more.

The following tutorial shows how to use Photomatix to create the HDR stills from a bracketing photography timelapse and how to assemble all the resulting HDR frames in a video using Gopro Studio.

And that’s all! Hope you found all this info useful. Please use the comments area if you want to share your experience.

As soon as I make more timelapse videos I’ll post them in this site and will write about them if there is something new or better about them.

Down here there is a button to share this article!

Be Sociable, Share!

    Leave a Reply

    One thought on “Trial and error with time lapse video

    • pegatinas para portatiles

      Muchas gracias por este post, tengo muchas ganas de seguir leyendo el resto
      de contenidos.