I’ve been thinking about editing for the last 5 years. Not just thinking, also editing. Trying out what works. You may have noticed different editing styles in the 65 videos I have posted to my YouTube channel.
I know that’s not a massive load of experience, but I’ll share what I’ve learned.
At the beginning I watched many how-to videos on YouTube. Tried to find out what is a good edit. I soon realized there isn’t an answer. All the tutorials just explain what is a J-cut, L-cut or jump-cut, how to do transitions, or some tricks with Adobe Premiere. Those are all good to know, but the most important question is, when you should use those tools. No-one explains that, at least in clear rules you could follow.
The reason is, I think, that editing is a creative art, where there is no right or wrong. People have different tastes and opinions. For example in movies the director and producer often battle on who’s got the final cut. The same goes to music, paintings, or arts in general – there is no rule book that everybody follows and agrees on. But still, some songs/paintings/movies/videos are popular and some are not. In some vague way people’s tastes point to the same direction. I guess it is just too difficult to put into rules.
I read In the Blink of an Eye by film editor Walter Murch. He explains you should cut when you would blink, physically or metaphorically, after understanding what you saw. For example, if you were watching a person who drones on and on about the weather, you would blink very quickly, or maybe even turn your head to watch something else. But if the person says he has cancer, you would stare at him in disbelief. So the context matters. The length of a cut should depend on the context, and it should replicate what an average person would watch if he was standing next to the scene.
That’s kind of deepish. In layman’s terms, you need to know what the viewer wants.
Knowing what the viewer wants is difficult. The viewer sees the video for the first time. You as a video creator have seen it once while filming it, and another 10-100 times while editing it. You can’t remove your memories and become the viewer. At least easily. Mr Murch says he hopes the director takes a 2 week vacation to forget everything, and then comes back to watch the edit. I find 2 weeks to be too little. 6 months is better. I’ve watched my own videos 6 months after releasing them and immediately noticed what I did wrong. Too late to fix those videos, but good for improving your craft.
Pacing, timing, rhythm, the length of each cut. I find this to be the most important. How much time the viewer needs for understanding each cut?
In movies, the trend has been towards faster pacing. I had to check from a study that my observation is correct. In the last 100 years average shot duration has dropped from 12 to 5 seconds. Try to watch 60s westerns, like the dollar trilogy, today. Very sluggish pace. Viewers today like fast edits.
You can see it in YouTube as well. Popular YouTube creators like MrBeast, PewDiePie and Dude Perfect use very fast edits. Maybe 1-2 seconds for each cut. Even in Lego scene, that I follow more closely, there are newcomers like TD BRICKS and Brick Science who use a very fast editing style. That’s a new style among Lego videos, and it attracts a lot of views at the moment. Kudos to those editors. Cutting 10 minutes of such a fast paced video is a lot of work.
Of course, fast edits are not always needed. Long form content is also popular in YouTube. Hours long podcasts and live streams with little or no editing get a lot of views.
When I started out, I remember fiddling with the edit and watching the result over and over again. Is this cut better 10 frames (0.4 sec) shorter or 10 frames longer? I could happily edit a fast paced version of the video, only to find the next day that it feels jerky and difficult to follow. It’s baffling to see your own taste for rhythm change from day to day.
It hasn’t got much easier after 5 years. I can edit more routinely the bulk of the work, but still the pacing requires a lot of polishing. I often watch the work dozens of times to feel out what is good.
Here is an example of a video I released in two version. First a 2.5 hour version with no cuts (no visible cuts, because it was recorded in multiple sessions, but edited into a seamless flow).
Here is the same material re-edited into a very fast rhythm. It is full of jump cuts. I edited most of the material using 5 frames for every click or snap where two Lego parts are connected. In some places the pacing slows down to let it breath a little bit. This is perhaps the fastest rhythms I’ve done. Somebody even commented the edit is too fast to understand whats happening. Rarely I get commented on the editing style.
Another example is my latest video, 20 Mechanical Principles Machine. It showcases different Lego devices one after another, like in a list. The key decision was to determine how long each device should be shown, and from how many camera angles. I knew I wouldn’t be able to evaluate it properly after I’ve started editing, because by then I’ve become too familiar to the material. So, before I started, I watched similar videos on YouTube (luckily I found several) and wrote down what I felt in a notepad file. What I noticed, is that the viewer needs a lot of time to examine each mechanism and how it works. Also I noticed cutting to a different camera angle distracts the viewer. The best version I found was this video that uses about 15 seconds for each device with no cuts. Very slow pacing. So I used that as a model for editing. I think it worked well.
Some of my videos are uploaded (without permission) to TikTok and get millions of views. Those uploads are often played in 1.5x or 2x speed. I take that as a hint that my videos could be edited faster. Often when you edit your own material, you have a tendency to over-emphasize everything. You had tens of hours creating the Lego build, set up lighting, and film it from different angles. You think the viewer wants to know each precious detail. That leads to slow pacing and over-usage of camera angles.
Structure, form, the arch from start to finish. That is another important editing decision in your video. In a way, that is the video, in the most abstract way.
When I started out, I watched other popular Lego Technic YouTube videos. Many of them were like this. Title: 6-speed Lego gearbox. Length: 2 minutes. The audio is filled with pop music that is annoying and distracting. There are long texts on-screen explaining how the gearbox works. The video has no beginning or an end, no story, just a static showcasing of the finished build from different angles. You know it lacks story when you can start watching the video half-way through and it is just the same. My primary feeling was, why is this a video? This would be better in a web article with text and a few images.
What I learned from that is 1) don’t use background music, 2) show things visually rather than explain it with onscreen text, 3) make the video into a story. I have followed those points with some variations throughout the last 5 years.
The story can be done at least two ways with Lego videos. The first is a building/making video. You start with nothing, build one step at a time, and end with the finished thing. The video structure is chronological, straightforward and simple. An example is a Lego Tank video.
The second way is a development video. You start with an aim in mind, e.g. make the most powerful Lego hoist. In the video you test, make improvements, test again, make more improvements, and so on. You may end up with something useful, or you may not, but that is not the point. The development is the point, not the end result. This structure is almost like the first one, chronological, but the progress is not so straightforward, as it bounces between testing and improving, and there is usually no clear end. You could always continue the development further. I did these types of video especially in the beginning, for example the Lego hoist video.
Besides those, I’ve done a few simple list structures. Just put scenes after another with a common heading. For example: All 19 Lego Breaking Moments So Far or Requested Lego Experiments 4. There is no story involved. Because of that, each piece must be short to keep it interesting.
A montage is another structure type I’ve used, but only as a part of a video, usually at the end to showcase the result of a build. It contains just short clips after another in no particular order. Kind of like a music video – that is why I combine it with background music. An example is from the end of Submarine 2.0 video.
Eye trace, eye movement, where the viewer is looking at between cuts. This becomes important in fast edits. At worst, I’ve seen fast paced videos where a text appears on the right side of the frame, then something else happens on the left side, then a new text appears on the right side, and so it bounces around the frame. The viewer cannot keep up. Every time the position of interest changes, the viewer needs some time to catch up.
I usually don’t have to mind eye trace, because I use steady camera and the Lego build stays at a fixed position. The point of interest (new parts added to the build) stays at a certain area in a frame, usually at the center.
But in other times I need to mind it. One example is when I hold the thing I’m building in my hand and it moves around. What I do in those situations, I use a wide angle shot, zoom in 10%-50% in the editor, and pan x and y position after every cut so that the thing stays in place between cuts. A side effect of this is that the background will jump around between cuts, but that is a lesser evil than the point of interest jumping around. An example of this edit is here (if you look at the upper right corner, you see how the background changes between cuts).
Part of minimizing the eye trace problems is in filming it properly. Lego speed build video creators know this very well. The construction has to stay in place while building it. Otherwise it will move around the frame, and those movements will become more pronounced and distracting when the footage is sped up. Optimally the only changing pixels on screen are the new parts added on the build (and perhaps your moving hands), everything else stays frozen. To achieve that, I often attach the build to the table using a double-sided tape.
Your hands may also draw unnecessary attention. When I hold the build with my left hand, while adding new parts with my right hand, I try to keep the left hand in the same place during the scene. Otherwise, jump cuts will make the left hand move around and distract the viewer.
Change in orientation is one more thing to consider. Optimally the Lego construction should stay in the same orientation. But if you must to rotate it, the act of rotation should be included in the edit. Otherwise the audience won’t understand what happened. An example from the obstacle car video where the car is rotated two times in the middle of building.
Continuity editing, consistency between cuts. Whether it is time, space, colors, lighting, or acoustics, if the continuity is broken, the result is confusing to the viewer.
In my simple videos continuity is fairly easy to do, but sometimes it requires effort. The biggest problem is with the continuity of the construction I am building. Whatever I have built earlier in the video, should be present later on. Many times I’ve shot a video half-way through, only to notice I built it wrong. If it is a small mistake, barely noticeable, I may leave it that way. If it is a larger mistake, the only way is to re-shoot all the sections where the mistake is present. That means lost time and effort. Because of those mistakes, I nowadays usually make a proper prototype and test it extensively, before filming anything.
Then there are smaller continuity checks to be made while editing. For example, I need to fix white balance and exposure between cuts. Otherwise, if the grey background changes in exposure while doing jump cuts, it will flicker and draw unwanted attention. Sometimes I’ve shot scenes using auto-exposure in the camera settings, and that footage always needs extra effort to even out the exposure variations.
Check the audio volume levels and effects between cuts. For example, if the level of background noise changes between cuts, it is distracting. Or if I cut to a different microphone that has brighter sound (Rode NT1 vs Sennheiser MKE600), it may be distracting, and I need to apply equalizer effect to match the frequencies.
When cutting to a close-up shot, if the change in distance is too small (for example 10% zoom), it will look like an error. Or if the change is too large (300% zoom), the viewer doesn’t understand what happened. In other words, the displacements should be neither subtle nor total. The same principle, called the 30-degree rule, applies to changes in camera angles. In terms of audio, a close-up shot should have higher volume level (or better yet, recorded using a mic in closer distance), as that is what the viewer expects. Audio should match what you see.
Sometimes continuity is deliberately broken. When cutting to an onboard camera, the soundscape, perspective, colors and lighting will all change, and that is fine, since the viewer understands it is different. The same goes to cuts between scenes, for example from a building phase to a testing phase. Continuity can be broken, since we have moved into a different mental space.
Different cuts I use
I mostly use jump cuts (cut forward in time without changing camera angle). That is because my footage is often Lego building scenes that are long and they need to be shortened. Jump cuts are naturally jarring and disorienting. Because of that, they are rarely used in films. But I think they work in Lego building scenes, because the changes between jump cuts are small (given that the camera is steady and the build stays in place while Lego parts are added). Here is a video full of jump cut building scenes. Each cut is about 16 frames long, and the snap sounds of Lego parts connecting are 6 frames in.
Sometimes, instead of jump cuts, I speed up the footage up to 4x or even 10x speed. This style was used often in my old videos, for example this Lego hoist video. That is one way to shorten the long building scenes. It is also easier to edit than jump cuts. But nowadays I think speed building has a bit rushed and unnatural feel, so I prefer jump cuts.
Dissolve is another cut I use often. Either Cross dissolve or Dip to Black. I usually put those to represent a transition from a scene to another, for example from a building phase to a testing phase (example). Dissolves take some time to finish, so you don’t want to use them often, as they will slow down pacing.
Cutting on action is another I have found useful when going from a full shot into a closeup, as the movement makes the cut easier to follow (example). One time I tried a cross cut between a cat and a bird, totally unrelated to the rest of the Lego video (example). A wipe transition I’ve used a few times (example) and push transition a few times (example).
I’ve learned a few effects and tricks to spice up the editing. Here are my picks.
I transformed gear train into a wireframe at the end of Googol video. A lot of people asked how I did that. It was done with a Adobe Premiere effect called Find Edges that comes with the software. Very easy to use effect where you can adjust the amount of blend and invert colors. Funny how I later came across an old AC/DC music video that uses the same type of effect with gears rotating in the background. What a coincidence!
I used an Old West flashback in the card gun video. It is done with a canyon background image, gunshot sound, wooden plank with text, and a white-edged highlight. I copied the idea from Placeboing’s video.
A blurred zoom transition was used in the card gun video. It is done with Transform -effect where you set shutter angle to 360 and then increase/decrease scale using keyframes. I copied that idea from a Dude Perfect video.
Animated dotted lines has been in a couple of videos, last time in Submarine 4.0. That is done with a Write-on effect by changing brush position over time with keyframes.
Many times I need to extend the frame of the shot. For example, if I want to add text to the side, but the framing is too tight for that. That can be fixed without re-shoot, if the background is uniform color at the edge. Continue the area with a single-color rectangle, or put the same video on another layer with different xy position. Below is an example from Submarine 3.0 with and without the patch up.
A few times I have used anime speed lines. Here is an example. It is a kind of cheesy cartoon-like trick, but I like it. I create it using this overlay video. Just lay it on top and from Effect Controls panel set Opacity -> Blend Mode -> Screen. I copied the idea from Davie504’s video.
One time I did a stop motion animation scene, just to try it out. Lego stop motion is a big genre on YouTube. There are many tutorials on that, but here is how I did it. I had only one minifig moving, so it was very simple to do. The walking movement was done in 4 steps: 1) left leg forward and right hand forward, 2) legs and hands down, 3) right leg forward and left hand forward, 4) legs and hands down. Exposure auto-adjustment was turned off from the camera (otherwise the edited result would flicker). A steady camera recorded video (you could also take pictures). I started moving the minifig, and took my hands out of the frame for a second after each step was finished. In the editor I cut 2 frames from each step and put them into a series. Simple as that. Here is the result. In some parts I combined stop motion and actual video together. That was done with masking the footage frame by frame.
A fast zoom to a closeup has been used a few times. For example in the beginning of the spinning camera video and rock lift video. It is done simply by keyframing scale and position. By the way, Kubrick did something similar in Dr Strangelove.
I generally don’t use music. The audio is only silence and Lego building noises. That is because a lot of other Lego Technic video creators fill the audio with music and it feels annoying and distracting.
But sometimes I like add a little bit feeling and personality with music to the end of the video. This often gets divided comments among viewers, some like it and some not. The audience retention graph shows a portion of the viewers stop watching when the music starts playing. Maybe I should stop using music completely, as people don’t seem to like it. But that feels a bit monotone and boring to me. I think the right kind of music at the right place is ok, I just need to learn how to use it.
My current opinion is that a very low-key music is best, almost boring music if you listen the piece by itself. When you combine that with the picture, it will add to a nice balance. Otherwise, if you select a normal good song, it will draw attention too much, and distract from the rest of the video.
One little part of editing is adding on-screen texts. I try to avoid texts as much as possible and let the visual imagery explain what happens. But sometimes a name of a device or gear ratio or something other information needs to be added. The font types I use are Tahoma, Leelawadee UI, Verdana, Arial, Eras. All of those are sans-serif fonts, simple and easy to read on a small screen.
To ensure readability, I try to put the text on an “empty” area that has uniform color. If there is no such an area available, I will add thick strokes or shadows around the letters. The color for the strokes is often picked from the background, so that it blends to the image. The color of the actual font is usually either black or white, whichever provides enough contrast. Sometimes I do a more stylistic choice, and pick the font color from whatever else appears on screen. Readability is always the most important aspect of texts. It is a good practice the watch the video on a cell phone to make sure the font size is big enough.
Timing, how long the text should be displayed? That is difficult to know while you are editing. Afterwards I often think I used too quick timing, and therefore the viewer doesn’t have enough to time to read and understand the texts.
Take care of your fingers
This is not related to editing, but anyways. My videos often include closeups of my hands while I build Lego creations. If I have dry and cracked skin and nails, those will show in the videos and distract viewers from the actual content. This is an area where I have improved over the years. My fingers looked pretty bad in the old videos, but much better in the new ones. What I learned to do, is keep up the moisture. Skin and nails look immediately smoother with some extra moisture. I may simply pour water to my fingers and wipe them clean, just before a closeup shot, or apply hand lotion an hour before shooting.
Building with Legos is of course hard on fingers and nails. I’ve learned not use nails for disassembling parts. Another thing I’ve tried out lately is a glass nail file to grind down edges of my nails, and cuticle oil. I think those help to keep the nails from cracking in the long run.
Lighting also makes skin look worse if it comes from a sharp angle and brings up all the grooves. Having more light from the direction of the camera helps in closeups.
After I’ve shot a scene, I immediately edit the material. I move the files to my editing software, and edit the footage before I start to shoot the next scene. I’m proud to say Akira Kurosawa operated the same way. He used to shoot during the day and edit in the evening. One benefit of this routine is the possibility to quickly re-shoot shots. If I have problems with focus, overblown exposure, hand blocking a key shot, or whatever, I can shoot it again without needing to spend time setting up everything to their original state.
I use Adobe Premiere Pro 2020 as my editing software. The first thing after I import files to Premire, is put them in a separate sequence. For me that is always Sequence 01 because I don’t care about naming them. That sequence is for collecting all the raw footage and synchronizing them. Synchronizing is needed to match audio and video, because I record audio with a separate device. If there are multiple video files, they also need to be synchronized. At best I have had 4 video files for an experiment captured from different camera angles.
Next I cut it. I move the new footage from Sequence 01 to the main sequence. I look through the footage to see how it looks. If it is a long scene, I may do a rough cut first to select which pieces I need, and then a fine cut to adjust the pacing.
Then I adjust color. I use Lumetri Color effect. I may put the effect directly to each video clip, or use an Adjustment layer on top to fix color for multiple clips at the same time. In the effect settings I usually fix white balance first, then exposure, and finally add contrast, saturation, and lightness for whites/shadows if needed.
I often watch Lumetri scope with RGB Parade while I adjust colors and exposure. I’m so accustomed to that tool, that I often tune white balance based on those graphs, without much looking at the actual image.
Audio is often the last thing I check. I may fix audio levels for individual clips to keep even level during the entire video. I use Parametric Equalizer, Dynamic Processing, and some other effects on the audio tracks. More about those in the previous post.
After everything is shot and cut, is the tuning phase. I watch the result multiple times in different days and improve on pacing, exposure, audio levels. I often watch the entire thing once on my cell phone to see how it looks on a small screen. Can you read onscreen text? How the audio sounds through cell phone speakers where there is no low frequencies? I also listen to the video once with head phones. Is the panning ok? Is there too much background noise or annoying clicks somewhere?
Looking at my old videos, I can see many moments I would edit differently today. Some are clear mistakes, and others perhaps a change in taste. Here are my picks.
Lego tank video was done entirely with background music and 70s style yellowish colors and film scratches. Especially the background music annoyed many viewers, and people complained about it in the comment section. I think they were right. Those edits were a distraction that added nothing to the content. I later released the same video again without music and filters.
Requested Lego Experiments 1 had a robot voice reading texts. I got that idea while watching Memenade that uses a robot voice to read out memes. Many of my viewers didn’t like that. Once again, the comment section was filled with complaints. So, I released the video again without voice-over.
100 Wheel Lego Vehicle. The structure is all over the place. First a long speed building scene with no cuts or closeups, kind of boring. Then a short stop motion animation that had little to do with the subject of the video. Then driving around with on-board camera, which was ok. At the end a Halloween scary reveal of the driver’s face. This video didn’t know what it was supposed to be.
Playing Card Lego Gun starts with showcasing the end result, and follows with the building and development phase. If I look at YouTube Analytics audience retention graph, I see a lot of viewers stop watching at 20 second point. I guess revealing the result at the beginning spoils the story.
Steel axle video has unnecessary switching between two camera angles in a few of the test scenes. It makes it more difficult to follow without adding anything new.
Also in steel axle test video, there is a short emotional scene smashing fists on the table. So fake and cringy.
Many of my videos has too fast pacing, especially with on-screen text. One example is in Lego tank video. The texts go away too quickly to understand them.
Gearbox video had in many places (example) a cut to a closeup and a jump cut at the same time. That feels slightly jarring. Today I would probably keep time continuity and only cut to a closeup.
The first two videos (example) were not white balanced properly. They have slightly reddish tint.
All of my first year videos (example) have an ugly stereo audio from using two microphones. Pronounced if you listen with headphones.
After listing mistakes, let’s pick good edits. Those videos that I’m proud of watching today.
Can Lego BREAK a Steel Axle? I like the pacing of this video. Even the usage of background music at the end fits well to the content.
Making Lego Car CROSS Gaps. A well balanced structure from start to finish, and a good fast pacing.
Engines 2 video. Another well balanced video.
Spinning a Lego Wheel Fast BY HAND
My theory for good videos
I’ve formed a theory for making good videos, an abstract rule that fits my way of thinking. That is based on information. As a video maker, I am trying to pass information into the heads of the viewers. The information includes how a Lego build was made, how it works, what were the problems, what principles were involved. Other things don’t matter, like stunning visual effects, getting people excited, showing off my skills, marketing my channel, promotion, or anything else. Just passing information. Ok, I might occasionally put in a nice visual shot for the sake of it, but at least 95% of the time it is just about passing information.
When you pass information, the main criteria are clarity and speed. You want to pass the information as accurately and without misunderstanding, and as quickly as possible. These criteria affect everything from filming, framing, lighting, to editing. First you frame the shot properly, with an angle that gets all the details the viewer needs to see. Usually that is an angled bird’s-eye view, as opposed to shots straight from the side. You try to light it so that every part is well lit and the subject looks 3-dimensional and with enough contrast. You remove any unnecessary stuff from the background to simplify information flow. You use a single colored background that is different color from the subject, to improve clarity and differentiation. You edit the material as fast a pacing as possible, while it still is comprehensible. You use eye tracing principles to make the cuts smooth and easy to follow. You cut out any unnecessary footage, derailed observations not related to the main idea, and other complexity from the video. You record noiseless clear audio where every sound is distinguishable, and don’t hide it under background music. In short, every decision should be determined by the information flow.
I think editing is the most important part in making videos. At least in my videos. Filming, lighting and audio are also important, but editing is number one. Not just because it takes the most time – about 25-50% of my video production is spent in editing. But also because editing has maybe the biggest effect on how good the video will be. Editing creates the rhythm, tone and excitement of the piece. Also, in a personal level, editing is the most fulfilling part of the process, because there you see your creation come to life.
Thanks for reading, and good luck with your own edits. Any comments, feel free to add them below.
Thanks for sharing, it’s very useful!
Thank you for this! Helps me think about how to edit both long and short form content.