Hey all first time poster. Hoping to get some feedback if upgrades to my rig will drastically reduce my editing time. Ive been using Shotcut for a few? years now.
From a few other posts I read it seems that ram, cpu and gpu are the questions people often ask:
CPU: AMD Ryzen 9 3950x 16 core 3800 Mhz 16 core, I did not overclock 2+ years old
GPU: 1080 ti 4+ years old.
Ram: 32 GB 2133 Mhz DDR 4 2+ years old.
Rig is heavily used.
I could go into much more detail but Im losing a lot of time in 3 main places. Dragging and dropping clips it will sometimes take up to 10-20 seconds to move the clip. Stuttering on playback (yes ive made adjustments using guides) and lastly export time (h.264 main profile quality moved to 100%, but sometimes when I accidently leave it on 55% the export time is the same). I believe im losing at least an hour on average I think 2 hours because of these problems.
As people that likely have much more experience in editing, will improvements to my hardware reduce my lost time during the editing process?
Im considering upgrading all 3 components with a 4090 as the main upgrade.
I’m far from an expert on those technical things, but from what I’ve often read here on the forum I’d like to mention two things (subject to confirmation by the more reliable forum members):
You don’t need to export at 100% quality. You shouldn’t be able to see a difference between, say, 65% quality and 100% quality.
Before spending hundreds and hundreds of dollars upgrading your CPU, GPU and RAM, maybe consider adding a SSD (or even better a SSD NVMe) if you don’t already have one. I’ve read they made a big difference for some Shotcut users.
Hey thanks for the initial response I only have 1 hard drive currently and its a Samsung 970 evo + 1 terabyte m.2 NVMe. And to your first point yea I agree with you I didn’t notice a quality difference and I fashion myself as someone who can see those differences. With that said I also didn’t see a difference in approximate encoding time, I normally walk away from the computer during encoding or focus on making my thumbnail so not sure if its making a difference in actual encoding time.
If you computer is slow then yes. If not, then no.
In Video-Editing class we learned that the biggest time-saver was to use keyboard shortcuts and not the mouse. This was for Adobe Premier and others and I don’t know how much this applies to Shotcut.
I’ve seen Professional Editors do this and they’re pretty fast.
I don’t know if there is a multi-machine Render queue for Shotcut, but on other systems you can distribute the Export load to idle machines around your home or work Network. There are even paid Cloud options for some systems. For me personally, I haven’t needed to use them. My machine runs acceptably fast.
Also, I break up longer projects and just export 30-seconds at a time max. I might do a full export every few days on a bigger job.
@david.lyon brings a good point here. Long projects with many tracks, clips and filters will slow down Shotcut. Chaptering your projects could help a lot.
@bentacular made and excellent tutorial on that subject a while ago:
And, just in case you’re not aware of this, using Proxy files and turning ON Preview Scaling will also help a lot.
We might be able to quickly save you some money on upgrades here. Shotcut does not use the GPU for frame processing. It is CPU only. Money would be wasted on the 4090 unless the plan is to use hardware encoding all the time, specifically AV1.
Overall, your rig looks pretty good already. Proxy and Preview Scaling would be the next optimization step, as already mentioned by @MusicalBox.
More details on the performance effects of other components:
More details on what Proxy and Preview Scaling bring to the table:
Hey all thanks for the responses! I read about proxy video sources in another thread. I currently use 360p preview sometimes the problem is when my gameplay is already far distance away the reduced quality sometimes its hard to see where I actually need to edit to show off what i’m looking for. For this I switch to 720p preview scaling to get what I need then switch back. It doesn’t help that WoW is already an ancient looking game. With this method or proxy/preview will I be able to switch to a higher resolution at some point to see what I am doing?
Separately with already using 360p preview scaling (without the proxy) I’m still having most of these issues, it helps a bit but recently its basically back how it was when I was using 720p and 720p is now unusable. Right now In order to get through a video I have to turn off some video tracks, toggle between video tracks. Because im doing so much moving back and forth just to get through the video I inevitably forget something and have to go back and encode multiple times all which usually take 15-20 minutes even for a 5-7 minute video. Chaptering might help this but even when I load like 2 segments the spot where the 2 segments meet start stuttering. If i start splitting every segment I might just end up in a different set of weeds (though I dont want to knock it before i try it) then the one I’m in. Its possible the card is just bad at this point Im starting to have in-game frame issues as well especially while streaming.
I dont mind doing these steps to speed things up. I am separately curious if buying a 4090 and upgrading ram would allow me to edit without downscaling the preview at all so I can work with the material as it will be presented to the audience. I do genuinely appreciate everyone’s help. I don’t currently have a project to work on so Ill have to get back on some of these solutions, hopefully by next week Ill have some content I can make.
Shotcut only uses the 4090 to encode the final video frame at the time of the final export. The 4090 would not be used to speed up building a frame for preview (as in, not used for applying filters or scaling). So no, the 4090 will not improve the computer’s responsiveness at all.
Stuttering in this case is happening because the gameplay video is encoded with LongGOP, which is not seek-friendly. LongGOP is a video trick that makes small file sizes possible, but it is computationally expensive to decompress, which is why the video stutters every time the playhead is moved. Creating proxies will remove the LongGOP and make for a much smoother editing experience. This is something that Preview Scaling alone cannot address.
In your case, perhaps the best solution is to start a new project, switch to 720p Preview Scaling, bring in all your media, and switch to proxy mode which will generate proxies at 720p since that’s the Preview Scale. Performance will probably be sufficient at that point. But if necessary, Preview Scaling could be dropped to 360p after proxies are made. This sequence of events would allow you to keep a 720p base to see what’s going on in finer detail areas when you need it.
These last few comments were amazing fam thank you for the advice. I might have some gameplay worth making a video out of tonight so maybe a video in the next few days and let you know how it goes. I appreciate all the help.
I think a lot has been said already, but I want to chime in because I have a much, much, muuuch worse setup than yours, and editing is seamless for me. Use proxy! Load the clips you need, grab a coffee, do something else, and wait till the proxy creating job is done.
Once that’s been completed, editing will be seamless with your beast of a machine. Exporting is another story, but I don’t think it’s worth it paying so much money for what I suspect would be marginal gains, also, that’s another part of your workflow that’s hands off, so changing your workflow to do something else while exporting would be more beneficial.
I have a 10years old AMD FX 8350 with AMD RX 570 and 16 GB of ram…I know it’s a silly setup, but I got the machine for 50 dollars, and I’m able to edit 4K 60fps GoPro videos without any issues or stuttering.
Good luck!
When I was using Nvidia GPU, exporting with hardware encoding made a huge difference. Of course, your GPU must support the chosen codec. If you’re exporting in non-supported codec, it will just resort to software encoding.
I switched to Radeon GPU recently, and found some issues that sometimes made hardware encoding slower than software encoding and the file size was several times larger. Not sure what’s going on there and that’s something i still need to figure out.
I used to use proxy files too. and yes, it did help a lot with making editing faster and easier, but I wasn’t happy with the preview since i’m looking at 720p image quality instead of UHD 4k. you can still edit, but the preview doesn’t accurately show what your final results will be.
However, I recently learned about intermediate codecs like ProRes and DNxHR. Some cameras can record in this format, or you can use an external recorder that records this format. Even if you can’t record in this format, you can use programs like ffmpeg to convert to this format. So, that’s what I do now. I have a script that runs ffmpeg on my MOV files and generates ProRes LT files. With ProRes LT, I can scrub through 4k footage with ease and I can see the actual image quality in preview. Everything feels snappy and faster in shotcut. It does take a little bit of extra time, but I just run the conversion in the background while I do other things ahead of when I will spend time on editing. The ProRes LT files are huge though, so you need to have a lot of storage during the editing. Once the project is done, I just delete the ProRes LT files and if I want to keep the original footage, I only save the original MOV files from camera which are smaller.
You already have a lot of good replies, but just to add my 2 cents (25 cents with inflation), I would suggest going with a faster processor with fewer cores. Like 5 GHz with 4 cores. Some apps simply can’t run on more than one core. Whereas something like a web server is easily threadable.
I have several machines that run shotcut and one of them is the same processor you have. Some tasks do take a while for sure, but in my case, its faster than my other machines that run atom processors at 1.5 GHz.
Lastly, its possible that using a linux OS might run faster than a Windows OS.
