Recording problems with a high end system - CPU Bottleneck?

[Dovahkitty]

Hi all,
I've been trying to start a YouTube let's play of Fallout New Vegas, with mods. However, I've tripped on a hurdle. I have a fairly high end system:
Intel Core i7 4770k @ 4.0GHz
Asus Geforce GTX 780 Ti w/ DirectCU II OC EVGA GTX 650 ACX Superclocked (dedicated PhysX)
Gigabyte GAZ87X-D3H
16GB Kingston HyperX Beast 1333MHz
Corsair H80i
Seagate Barracuda 1TB 7200rpm Hard Drive
Samsung 840 EVO 250GB SSD
WD Caviar Green 1TB 7200rpm Hard Drive
XFX PRO 650W PSU

BUT: when I try to record Fallout New Vegas (which only uses two cores), I get a huge drop in framerate (20fps ish) with Bandicam (I've tried a variety of codecs), DXTory and Mirillis Action. As I only have one rig, my hardware solution still hits performance - I can't do something like Gopher's set up where I have two machines, one for the capture, one for the gaming. It's an AverMedia ExtremeCap U3 (requires a USB 3.0 port). Th only machines I can use with that are my gaming rig (above) and my laptop - which is far too weak to decode even 720p60 on YouTube, let alone 1080p60 60Mbps video. 1080p60 is very important to me, and I'm not sure if I can achieve it.

Any suggestions on how to improve this?

Thank you in advance,
Dovahkitty

 

[Frag Maniac]

Try using DxTory's Distribution Writing feature.

To do this you need to set output type to RawCap instead of AVI, then you can designate two or more drives as the folder locations for the capture. What this does is use a kind of virtual RAID whereby the drives simultaneous write RawCap files in each of their folders, boosting performance.

The only caveats are, you have to use the DxTory codec, and after capture you have to use DxTory's RawCapConvert tool to construct the AVI from the RawCap files.

The DxTory codec writes pretty large files though, so I recommend starting with a test sample in Low quality, which is probably close to ShadowPlay's max quality.

You can also set RawCapConvert to delete the RawCap files automatically after the AVI is constructed. I recommend this because there's no point keeping the RawCap files.

Speaking of ShadowPlay though, you have that option as well with a 780 Ti, and it's by far THE most convenient since it has 20 min cache recording capability at 1080p. This means you can replay checkpoints until you're satisfied with the outcome, and THEN hit the Shadow record key to capture the last 20 min of gameplay, without wasted drive space or trial and error recording.

In fact I would try ShadowPlay first, since it makes use of hardware encoding by accessing the H264 encoder built into the GPU, and is thus very well optimized as there's no software lag. It basically accesses architecture on the chip that otherwise sits dormant, and it's not recording WHILE you're playing, but rather AFTER you end a 20 min session.

You can as well set ShadowPlay to record manually in the background though, and in that case, there's no 20 min limit. It's also pretty well optimized even in manual record mode.

https://www.youtube.com/watch?v=W0KqsTa_HpM

https://www.youtube.com/watch?v=w7ysCP8k9iI

 

[Dovahkitty]

Try using DxTory's Distribution Writing feature.

To do this you need to set output type to RawCap instead of AVI, then you can designate two or more drives as the folder locations for the capture. What this does is use a kind of virtual RAID whereby the drives simultaneous write RawCap files in each of their folders, boosting performance.

The only caveats are, you have to use the DxTory codec, and after capture you have to use DxTory's RawCapConvert tool to construct the AVI from the RawCap files.

The DxTory codec writes pretty large files though, so I recommend starting with a test sample in Low quality, which is probably close to ShadowPlay's max quality.

You can also set RawCapConvert to delete the RawCap files automatically after the AVI is constructed. I recommend this because there's no point keeping the RawCap files.

Speaking of ShadowPlay though, you have that option as well with a 780 Ti, and it's by far THE most convenient since it has 20 min cache recording capability at 1080p. This means you can replay checkpoints until you're satisfied with the outcome, and THEN hit the Shadow record key to capture the last 20 min of gameplay, without wasted drive space or trial and error recording.

In fact I would try ShadowPlay first, since it makes use of hardware encoding by accessing the H264 encoder built into the GPU, and is thus very well optimized as there's no software lag. It basically accesses architecture on the chip that otherwise sits dormant, and it's not recording WHILE you're playing, but rather AFTER you end a 20 min session.

You can as well set ShadowPlay to record manually in the background though, and in that case, there's no 20 min limit. It's also pretty well optimized even in manual record mode.

https://www.youtube.com/watch?v=W0KqsTa_HpM

https://www.youtube.com/watch?v=w7ysCP8k9iI

Even the DXTory low codec hits my fps pretty badly, and Shadowplay means that I can't split the audio tracks, so that's out. 🙁 Also, my playing sessions are longer than 20 minutes - I can bash out 40 minutes of video without a problem, and that, to me, is a good let's play episode size. The problem I had with DXTory was not that the game FPS was low, but the written file fps was something around 9 or 10fps. Would this fix that?

 

[Frag Maniac]

And you want to split audio tracks why? I understand that once all audio is recorded on one track it becomes impossible to remix it afterward, but there ARE things you can do before recording to account for things like mic sensitivity, etc. For example adjusting mic volume separately from gameplay volume.

Also, as I said ShadowPlay can be put in manual record mode, and still runs at decent FPS that way. Since you're comparing it to a tool that cannot even do cached recording, I would think that would make it a plausible option if you can tweak your audio setup to an acceptable level.

If not, As I said DxTory can be used in Distribution Writing mode. Have you even tried it that way?

 

[Dovahkitty]

And you want to split audio tracks why? I understand that once all audio is recorded on one track it becomes impossible to remix it afterward, but there ARE things you can do before recording to account for things like mic sensitivity, etc. For example adjusting mic volume separately from gameplay volume.

Also, as I said ShadowPlay can be put in manual record mode, and still runs at decent FPS that way. Since you're comparing it to a tool that cannot even do cached recording, I would think that would make it a plausible option if you can tweak your audio setup to an acceptable level.

If not, As I said DxTory can be used in Distribution Writing mode. Have you even tried it that way?

My microphone is very quiet, so I need to almost triple the volume in post. Also, say a member of the family opens or closes a door outside, or something falls, etc. I need to cut that out. It's pretty much essential. But thank you for the suggestion. I will give DxTory a go, many thanks.

 

[Frag Maniac]

My microphone is very quiet, so I need to almost triple the volume in post. Also, say a member of the family opens or closes a door outside, or something falls, etc. I need to cut that out. It's pretty much essential. But thank you for the suggestion. I will give DxTory a go, many thanks.

What brand/model of mic or headset are you using?

If you don't have a noise cancelling mic and it's a short stick vs mini type, try helping it a bit by rigging something around it to channel your voice-only into it. You'll want to do that with something like open cell foam to avoid a tube echo effect though.

I would suggest something like 1/4" sheet foam formed into a slight cone shape and rubber banded, taped, or zip tied around the mic. The foam can be found at local shops that sell foam (ask about scrap bins), or online at places like Amazon.

You don't need high grade foam either. The plain yellowish white stuff is fine, and is commonly used on mics for things like blocking out wind and other noise. A cone made of the stuff might suitably block exterior noise and channel your voice into it better.

If it's just a headset mini boom mic, you still might be able to effectively channel your voice into it better with a smaller, thinner piece of foam fashioned into a spoon shape. Keep in mind though that boom position and focusing on directing your voice straight to the mic can make a HUGE difference. I often have to ask people at certain businesses I call to speak into the receiver better, and sometimes it's just due to poor boom mic position.

Don't underestimate small audio wave diverters though, they can make a big difference. There's a rabbi I saw on Shark Tank who's now making big bucks off of an invention that is basically a small clip on spoon-like piece of plastic that directs sound from the tiny speakers on portable devices like phones and tablets, making what you hear much louder and more distinct.

Other than that, the room you do it in is key. Avoid open doors and windows if you can. Also, avoid smooth surfaces where possible, like tossing a rug over bare floor, closing drapes over windows,etc. The bare surfaces reflect sound and affect recording quality.

Not many software capture tools can do dual audio stream recording like DxTory though, which starts paying dividends on down the road. High end noise cancelling mics can be expensive.

 

[Dovahkitty]

My microphone is very quiet, so I need to almost triple the volume in post. Also, say a member of the family opens or closes a door outside, or something falls, etc. I need to cut that out. It's pretty much essential. But thank you for the suggestion. I will give DxTory a go, many thanks.

What brand/model of mic or headset are you using?

If you don't have a noise cancelling mic and it's a short stick vs mini type, try helping it a bit by rigging something around it to channel your voice-only into it. You'll want to do that with something like open cell foam to avoid a tube echo effect though.

I would suggest something like 1/4" sheet foam formed into a slight cone shape and rubber banded, taped, or zip tied around the mic. The foam can be found at local shops that sell foam (ask about scrap bins), or online at places like Amazon.

You don't need high grade foam either. The plain yellowish white stuff is fine, and is commonly used on mics for things like blocking out wind and other noise. A cone made of the stuff might suitably block exterior noise and channel your voice into it better.

If it's just a headset mini boom mic, you still might be able to effectively channel your voice into it better with a smaller, thinner piece of foam fashioned into a spoon shape. Keep in mind though that boom position and focusing on directing your voice straight to the mic can make a HUGE difference. I often have to ask people at certain businesses I call to speak into the receiver better, and sometimes it's just due to poor boom mic position.

Don't underestimate small audio wave diverters though, they can make a big difference. There's a rabbi I saw on Shark Tank who's now making big bucks off of an invention that is basically a small clip on spoon-like piece of plastic that directs sound from the tiny speakers on portable devices like phones and tablets, making what you hear much louder and more distinct.

Other than that, the room you do it in is key. Avoid open doors and windows if you can. Also, avoid smooth surfaces where possible, like tossing a rug over bare floor, closing drapes over windows,etc. The bare surfaces reflect sound and affect recording quality.

Not many software capture tools can do dual audio stream recording like DxTory though, which starts paying dividends on down the road. High end noise cancelling mics can be expensive.

My apologies for the late reply - it got late here in England. I have a Behringer C1-USB mic on a moveable arm, and I set about 4 meters away from any windows. I do have a pop filter as well. Background noise isn't the problem, but "accidental" sounds.

 

[Frag Maniac]

I have a Behringer C1-USB mic on a moveable arm, and I set about 4 meters away from any windows. I do have a pop filter as well. Background noise isn't the problem, but "accidental" sounds.

Well, technically accidental sounds are still in the background. A conical foam shroud still applies to minimizing such noises, while making what it picks up more directional too, if shaped properly.

Here's a more elaborate version of what I'm talking about, but much bulkier and more expensive than necessary, and probably wouldn't serve as well for directional sound as what I suggested.

https://reverb.com/item/929896-portable-microphone-studio-voice-booth-isolation-box?_aid=pla&pla=1&gclid=CjwKEAjwovytBRCdxtyKqfL5nUISJACaugG1ISELokpxKVsOfqqJAkaYxTyl8rHCQl1Wk_obsXPoeRoC50Pw_wcB

 

[Dovahkitty]

I have a Behringer C1-USB mic on a moveable arm, and I set about 4 meters away from any windows. I do have a pop filter as well. Background noise isn't the problem, but "accidental" sounds.

Well, technically accidental sounds are still in the background. A conical foam shroud still applies to minimizing such noises, while making what it picks up more directional too, if shaped properly.

Here's a more elaborate version of what I'm talking about, but much bulkier and more expensive than necessary, and probably wouldn't serve as well for directional sound as what I suggested.

https://reverb.com/item/929896-portable-microphone-studio-voice-booth-isolation-box?_aid=pla&pla=1&gclid=CjwKEAjwovytBRCdxtyKqfL5nUISJACaugG1ISELokpxKVsOfqqJAkaYxTyl8rHCQl1Wk_obsXPoeRoC50Pw_wcB

I see. I will look into those.

 

[techienumber1]

I just wonder why people who say the word bottle neck dont understand electronic's where there is chokes to capacitors there is no bottleneck because they are all smoothed out with things like these I know I have been building circuits like TVs and other electronic devices bottleneck just doesnt happen
until you know what you are talking about keep quiet and leave it to us guys who really know the stuff