I don't usually publish in English, but I have translated my original article to help as many people as possible. This issue is overlooked for some, but it is a source of headaches, frustration and obssession for others: "why does my gameplay recording stutter every few seconds?" Let's try to fix it!
Oh, by the way, Google Translation sucks, I need time to check and improve this guide with proper English. Any suggestions to AlexeiBono (Telegram), thank you.
I had prepared a long introduction and suddenly I felt like a mediocre SEO blogger, bursting my article with irrelevant information, like someone desperately looking for visitors. To me, it's one of the reasons why the modern internet is uninspiring and sucks so much: basic content about many topics at once without delving deep into any of those subjects. I will do my best to focus on a single topic, I don't want to be like them.
I would simply like to remind you that there are a thousand different ways to preserve the classic video game and almost none of them consists of setting up a private "museum" at home (If you build a museum, allow everyone to visit it). At the end of the 90s, I was fascinated by capturing videos with the same card that I used to hack Canal+, but at that time it was unthinkable to host them on the internet, only big sites like IGN or GameSpot could afford to publish some video clips with their news and reviews. I saw Shenmue for the first time on IGN, what memories!
But now, we have so many ways to do it right, I wonder why we're not doing it. Seriously, we're not. And this is not a criticism of others, in fact, it self-criticism. So at this point, I'm going to do my best to capture with the highest possible quality while respecting the original experience as much as possible. As I always say, if you don't put yourself in front of a CRT screen, you don't get the pure 100% original experience; that's how it is. But with a good scaler, a suitable monitor, a decent capture card and a valid configuration of each element, we can capture the signal of an old system with very high fidelity and the same speed, smoothness and original response.
Doing it correctly, the result is amazing and we can create a more faithful video library than the usual emulators.
My scaler GBS-Control
I use a scaler to convert the old signals into something my modern capture card and monitor can understand. Mine is a GBS-Control made following the instructions of Rama, author of the project. On my channel LV-481 I have a video about its construction in... Spanish. Sorry.
Other scalers I recommend: GBS-Control, OSSC and Retrotink5x.
My capture card: Elgato HD60 X
You will need a capture card compatible with your scaler, or rather, compatible with the out-of-range frequencies that they sometimes use. Do you know what is the problem? That the old consoles did not emit at 60hz, but in approximate figures. To give you an idea, I have a Super Nintendo emitting at 59.55hz, a Mega Drive at 59.46hz (I think), another at 59.92hz... And those signals are a bit far from 60hz, most capture cameras don't like them.
My previous device was the Avermedia Live Gamer Mini that did not accept frequencies lower than 59.80hz, I had to modify the vertical synchronization of my scaler (breaking the smoothness of the gameplay) to be able to capture. I assure you this is not what you want. On the other hand, Elgato's HD60 X is accepting frequencies slightly far from 60hz, although I cannot guarantee that all consoles and scalers will work. It is a very complex matter, do not give up.
How to choose capture? Amazon. Test and return, test and return until you get it right. I hate doing it like this but we have no choice, it doesn't seem that the best-known brands have qualified personnel for issues as technical as ours, they are trained in what today's gamer demands: live streaming of current games.
When you choose a capture card, if you are looking for the highest quality, don't let it take care of the final encoding. If you want a clean image, compression will be left to your computer. A capture card like mine sends the signal to the computer without compression and it is your OBS configuration that will determine the result. Well, technically, there is minimal compression, we go from a full 4:4:4 signal to a 4:2:0 subsampling, which means that some data in the image is ignored to simplify the flow of data.
But don't worry, a YUY2 4:2:0 capture still falls under the category of "lossless", the quality is excellent. I'm unable to find differences between what I see while playing and what the computer is capturing.
Elgato HD60 X: https://amzn.to/3p6DevL
The horrible stuttering that dominates everything
The "microcuts" or "stuttering" are the visible result of desynchronization, they occur because the source emits at a different number of frames than the value that the screen/capture card reproduces that content. As I said, they're not completely synchronized. For example, if my escaler is sending 59hz and I'm capturing at 60fps, every second there will be a frame skipping, obvious even to untrained eyes.
In a quick search, the first gameplay I found on YouTube suffers from this problem. It is no coincidence, the entire internet is full of videos like this:
Can you see how poorly the map before the first level scrolls to the left? Do you notice all those tiny "jumps" in the image? I would swear that on my console, that scrolling is smooth as butter. If we have the knowledge, wouldn't it be great to capture videos that really are a true reflection of how these machines worked? That's what I'm trying to do, I'm helping you to achieve the same quality of my videos. I think that watching videos of old games should be true joy and a neverending pleasure.
How to capture retro consoles without microcuts/stutter
The solution is to get your console, your scaler, your screen and also OBS all set to same hertz and frame rate, all of them working in harmony!
This is how I do it.
Console and scaler: with the GBS-Control nothing has to be configured. I enter the menu, choose a resolution and the frequency that the scaler receives from the console (source) and the one it sends (new out) to the capture card appears on screen. In this case, the output frequency is exactly what the source is producing (sometimes there is a slight variation): 59.92141hz.
OBS: I go to SETTINGS, look for the VIDEO section and add the previous value as frame rate. But there is a trick, they won't let me select it, or type it, I have to enter it as a fraction. Instead of spending the afternoon trying random divisions that give results close enough to 59.92141, I'm going to resort to something I haven't seen anyone do (I don't know why, it's the simplest thing in the world).
As a numerator, I enter the frequency at which my scaler is working, but without the point: 599214. I am missing a digit, it doesn't fit, no problem, that loss will be negligible.
As denominator, I simply enter 10000.
Why isn't there an easier way to enter the framerate you want to capture at? jaslñdkfhoasfhasñhf DAMN I DON'T KNOW, OBS WORKS LIKE THAT.
What will I get when OBS divides 599214 by 10000? According to the VLC video player, I have a capture at 59.9214000 frames per second, which is almost, practically, the same value that the scaler and console output at. Perfect!
Screen (part one): using the CRU (Custom Resolution Utility) tool or from my graphics card's control panel, I create a new resolution with a refresh rate as close as possible to, again, the console's and scaler refresh rate. In my case the closest number I can get fluctuates between 59.921730 and 59.92185hz or so.
I wish I could get closer, but neither the CRU nor the NVIDIA panel allow me to enter enough decimal to be that precise. The result is a bit of a random number, due (I imagine) to spikes and changes in the state of the CPU and GPU, fluctuations and internal maths that I don't understand.
Visit displayhz.com to check the refresh rate of your screen, let the counter run for 10-15 minutes to stabilize. Depending on the tasks you are performing on your computer, that number may vary slightly. Try to create an environment similar to the recording one, leaving everything open that you will need and with the same power settings.
That 59.92174hz value isn't bad at all, but it misses my target (scaler and console frequency) by about 0.00033hz, or even a bit more if there's any fluctuation (they seem completely normal).
Will such a small difference between scaler and screen be noticeable? No, the image should be perfectly smooth, but if my calculations are correct (I could be wrong ha ha ha), I would say that every 50.5 minutes there will be a sync skip, some kind of stutter.
How do I know when a stutter will occur? I calculate the hertz difference between my source (scaler or console) and my monitor: 59.92174 - 59.92141 = 0.00033hz.
Now, I divide 1 by that difference and I'll know how many seconds will pass until the next stutter: 1 / 0.00033 = 3030 seconds.
And 3030 divided by the 60 seconds that 1 minute has, is 50.5. The stutter will appear every 50.5 minutes.
I could live with it, but it can be fine tuned. Let's go!
Screen (part two): ok, I have left my screen at a frequency close but not definitive. Now, I open RTSS (RivaTuner Statistics Server), add OBS and limit its framerate, that's right, to 59.92141hz, the frequency of my consoles/scaler.
And so I solve the problem that brings the internet headlong to specifically capture classic consoles without microcuts or stuttering under Windows 11. I no longer have Apple computers to test, I can't help anymore.
Is it possible to skip the first step of the screen, leave it at any frequency and just limit OBS framerate with RTSS? In my case, no. If I leave my monitor at 144, 120 or even 60hz, OBS preview and capture come out less smooth. On the other hand, if I reduce the refresh rate of my monitor to a value closer to the console I'm capturing, and then limit OBS with RTSS, I get super smooth scroll comparable to playing on a tube TV. Those two steps make the difference, at least in my case.
What will happen when you edit or upload those videos captured at such weird frame rates? It is not known. I also don't know exactly what other people will see on screens with different frequencies. I would tell you that if you love old video games and you like to watch games captured at 60 frames, put your computer screen at 60hz. However, although I intend to upload everything to YouTube, I will also store my raw (unaltered) recordings in the cloud for anyone to view or use them.
Could this guide be used to broadcast live without microcuts? Yes and no, it depends on what you want. If we refer to old systems, I would say that it is impossible. Platforms like Twitch have their transmissions fixed at 30 or 60 exact frames and, unfortunately, the old consoles do not reach that value with enough precision (you have already seen the Mega Drive that I use as an example with its 59.92141hz). But if you're streaming PC games and modern consoles, you can probably set everything to 60hz, or so close to 60, it could take hours before there's a sync failure.
Remember that if you make a purchase through my links, you help me finance these crazy things that I get myself into. Because telling you my solution is a long time writing, but learning, that's a lifetime of messing around.
Elgato HD60 X: https://amzn.to/3p6DevL
The HD60 X seems a fabulous capture card, but I don't know to what extent it can be profitable for the average hobbyist, I would be very sorry to create a need for you. What I can tell you is that I wouldn't even pay the price on its official website. Be patient, on Amazon it fluctuates and it is normal to find it for 30 euros less from one day to the next.
Also look at second-hand and refurbished offers, at some point you will find one that is a little more affordable. There are streamers constantly renewing their equipment, the world of gaming -unfortunately- has become an addiction for being up to date, there will always be a good capture card that someone has replaced with another one with practically the same specs. This is how this industry works, news with a dropper. They're all like that, I guess.