After surveying the forums and experimenting with different ways to set levels, it seems clear that there’s no real consensus on how to optimize. I think I came up with a way to get consistent results that get the target device and capture levels as close as possible. I haven’t solved the difference in gain levels issue (QC seems to sample less gain than the target device) that has been widely noted, so if anyone has any suggestions for that specific problem, please share them here. Gaining the target device up a bit prior to capture seems to help.
Here’s my approach. See if it works for you.
Use a looper pedal and record yourself playing at the top of your dynamic range. Try hitting big 1st position open chords as hard as you can.
Using your recorded guitar, set the Inst Level as high as possible without inducing clipping as mentioned in the manual. Don’t be afraid to push the level. Right now I’m capturing my Bad Bod Boost and the sweet spot is all the way up around +8db (meter occasionally blips orange).
Use Auto-Set to set In 1 Level. Let your looper play 3 or 4 times (as many as it takes) to find peak loudness. The level will bounce around quite a bit, which is ok. Just watch the levels closely and figure out where it tops out. Then set the level to about -1db or so lower than that peak level. In my case with the Bad Bob Boost cranked to about 3 O’Clock, it seems to top out around -9.6db.
You’ll notice that the higher the output of your target device, the lower the In 1 Level will be. Just run this routine each time your capture and should get a pretty consistent result.
Try it out and let me know if your results improve!
This refers to level settings only, I think electrical matching (impedance) is an important factor too, here it is set to 470Kohm, (imput level makes a huge difference, you can make a Deluxe sound like a high gain amp) first thing I listen for is the bass response when comparing a capture to the reference, as soon as I hear the bass strings lose their hair, like a flat wound string I reject the capture, I don’t even have to compare, I got different options to capture, I have a Powerstation1, 2Notes live, a BSS DI box and a Marshall SE 100, as far as I know the first reactance load box I haven’t used lately, sometimes a capture on the PS will give that flat wound string thing, doing the same thing on 2Notes live will yield better results and visa versa, some amps are easy to capture some don’t, like I have a boutique amp FXLR, a fantastic amp that I only can capture with a mic, some of my capturing devices come close but not close enough for me regarding this amp, been trying different impedances on in1, it makes a difference but I’m not too familiar with the differences, I have the impression that capturing via XLR yields better results, like the BSS DI.
I have a few wishes though, 48V phantom power off by default, so you won’t fry a ribbon mike, a pilot tone to set the levels would be welcome, maybe white noise to accurately position a mike, I would love reversed tape delay, my 2 cents.
Let’s discuss this some more. Maybe @jamsden can also chime in since he had some similar findings in the following thread: Clipping outputs?
In this video NDSP uses the autolevel function to set the input 1 level for capturing a pedal. After some attempts, I’m really beginning to question if this is the correct way for capturing pedals. Autolevel seems to maximize the level without clipping. However, it seems that, by doing this, the capture ends up with a much higher level than the pedal. Comparing capture vs. real pedal before an amp block results in quite a difference of perceived total gain.
Rabea is capturing his pedals in a similar way, maxing the input 1 level without clipping. However, in his settings of the resulting capture (I’m referring to this moment in his video), you can see the level being set to around -14 dB in order to match the output of the capture to the output of the real pedal.
So my theory goes like this and it would be interesting to hear your and maybe NDSPs opinions on this:
If we capture an amp then it makes sense to max the input 1 level in order to have consistent output level of different amp captures. However, if we capture pedals, maxing input 1 level results in all the pedal captures having the same output level. This is rarely the case when comparing real pedals and especially makes no sense when making different captures of the same pedal but with different level settings. So in order to capture a pedal without needing to heavily adjust the capture output afterwards it makes most sense to me leaving the input 1 level at 0 dB for each pedal.
What do you think?
See Capture Techniques - Quad Cortex Wiki for my recommendations. Basically, it comes down to how to get the gain (saturation) and level of a capture block to properly reproduce the transfer function of the captured device. That is, the pedal in an effect loop in QC should behave the same way as a capture of that pedal in the same place in your QC signal chain. Both should have the same gain/saturation and the same level. This is how I test my pedal captures.
For pedals, I find using input at 0dB makes sure I’m driving the pedal the same way my guitar would when adjusting the pedal before the capture. Generally, I set up the pedal in a realistic signal chain and not during the capture setup, then just don’t change the pedal settings during the capture. The input level has no effect on the capture, it only impacts what you hear during the capture setup. So for pedal captures, I don’t generally even bother plugging in a guitar or setting the input during a capture, but if I did, it would be 0dB.
To get the capture to have the same level as the pedal, I set the capture in also to 0dB. I find this results in capture blocks of pedals that closely match what the pedal would do inserted at the same position in the signal path.
For amps it’s a bit more complicated because of headroom considerations. Clean captures will need more headroom than overdriven captures. What I did was explore a few typical clean and distorted amps in QC to see what their output levels were with the block level set to 0dB. What I found was that clean amps are somewhere around -12 dB while overdriven amps were around -6dB. That’s probably to 1) give more or less consistent levels between amps and 2) to account for the wider dynamic range of a clean amp.
Based on this, I set the capture level at -12dB when capturing a clean amp and -6dB when capturing an overdriven amp. That seems to result in a capture block that behaves consistently with other captures I might create and is similar to QC amp blocks.
Thanks for confirmation and for the hint to the wiki page. This well written explanation can’t be shared enough. I’m wondering though, why NDSP is suggesting oviously less ideal settings for pedal capture in their instruction video. Maybe there’s a signal to noise benefit when using the max range of input 1 and then reducing the output of the capture block by the same dB amount afterwards?
My guess would be that capture in level determines the 0dB level of the resulting capture block. It might improve signal-to-noise to keep this hotter, but I doubt that is a significant issue. Another possibility is that it’s easier to turn the capture block level down if it has too much level than it is to turn it up to recover missing level. The capture block level control has no impact on the block’s tone, so maybe it’s not really that much of an issue as long as the capture has enough gain.
Another possibility is that using auto capture in leveling would produce similar capture block output levels. But I don’t think consistency is necessarily what we’re looking for here. We want the capture to behave like the capture device. So for example if you have low gain and high level on an overdrive pedal for a boost, you don’t want the capture to reduce or change that level, you want it to reproduce it.
I do think having consistent levels in captures of similar devices would be convenient and simplify patch editing. And there seems to be little motivation to dance around the edge of digital clipping on capture input.
Also, the the capture signal level is always the same, regardless of what you set the out level to. Only the return level is actually important for the capture itself. Next time I do captures I’m going to add a volume control between the QC and the target so I can actually adjust the capture signal level and see how that affects the capture quality.
In this case the capture is most likely to sound like the target played with the volume pedal in front, I guess.
I’d recommend experimenting with the input level set in the global I/O window as well as the ‘Input 1’ level shown in the capture window. I believe that most people set their inputs too low & that’s why their captures tend to end up with less gain than the original device.
For me my first set of captures had too much gain, I then realised that when I’d initially setup the levels in the I/O window I had set the level for Input 1 too high, I therefore reduced said input level accordingly & those same captures then had slightly too little gain (inline with what a lot of people complain about). I then redid the captures with the input level set to my newly chosen level & replicated the same level for the capture return (‘Input 1’ in the capture window)
this time the gain levels are identical to the original device (still some slight EQ differences but that’s not to do with levels).
It is correct that the instrument input control shown in the capture window is totally irrelevant to the capture; however, both your global I/O settings & the capture return level are extremely important in my experience - it’s worth the effort to get these right before creating too many presets as once you have it dialled in correctly then all of your captures will be as accurate as possible in terms of gain levels.
I think it’s more a tool when dialing in the device to be captured. For example, if you have your global input level set other than 0dB, setting the input control to a similar value in the capture window will make sure, you can audition the device to be captured with a similar input level that the final capture will be subject to in your patches.
does changing the actual global input level affect the send level for capture testing? Seems like there wouldn’t be a need to correct for it if it just played your signal through using the global IO input level. Would be much more useful if it actually changed the signal send level for capture
I don’t think so and was not trying to state that. The capture mode seems to be a completely independent mode from whatever is set in the “play”-mode and I think it’s important and correct to be so in order to have a consistent baseline for capturing regardless of user preferences on different QC units. If the global input level would have any influence on capturing, I would expect to be able to set it in the actual capture mode but the I/O screen is not accessible in this mode.
My point was that INST level in capture mode might help you when you’re dialing in e.g. an amp being wired up in this mode before capturing. If you are used to set global input level other than 0dB, you could set INST level in capture mode to the identical value and then dial in your amp based on the resulting signal. This should provide a similar perceived gain later on in the patch.
However, I’m beginning to doubt whether this is a good practice at all. It would probably be most consistent for all when devices to capture were dialed in based on INST level setting of 0dB even if the actual signal of the connected instrument doesn’t peak at 0dB with this setting.
Totally agree with that last part. I personally feel that changing the instrument level in the capture window just confuses matters & the best results are achieved if this is just left at 0.0db. I feel that the optimal approach is to set target pieces of gear the way they would be set when you actually use them & to then aim to set your global input on the QC so that your captures; when inserted in the grid & actually in use, perform as closely as possible to the original pieces of gear (in my experience this requires a significant boost as if left at 0.0db the signal going into the grid is much weaker/quieter than when one plugs their guitar straight into the target gear without the QC involved - although this seems to vary depending on the devices captured so it really is about finding the compromise that suits you best & then being willing to tweak certain captures from there).
How things sound in the capture window are irrelevant to me as this is not how/where the captures are actually used. I feel that the only thing worth checking here is the input level from the capture return & whether or not the ‘reference’ & ‘cortex’ tones match up; the actual tone itself is not important at this stage as that part will need to be reviewed & tweaked as needed once in the grid (I use an A/B switcher for this so that I can easily & quickly compare it directly with my guitar going straight into the original piece of gear).
This link from above was very helpful - I thought starting with a DI device capture would make life easier than sorting through amp mic’ing, but obviously I was wrong… :-).
My first device was the original Rockman Distortion Generator as there’s a lot of variations to capture and there weren’t any in the cloud yet. Levels and gain were such a hodgepodge disaster I actually ran it through a Milkman and an old Boogie cabinet (with a new celestion after the original EV died) and captured it mic’d to see how that worked.
You can see them all in my profile (oandrew99) online but after reading the above link I redid the HVY profile direct and it seemed to work very well. There are four presets up there in my cloud - the Rockman DGS are the speaker captures, and the DG-D is the direct capture. The direct is truer to the original I think, and a little less crispy - the Rockman is a little dark sounding and I’ve left it that way, but it adjusts well with the tone controls in the neural capture.
The Distortion Generator had four levels of “Sustain” (30/40/50/60) and then three levels of “Distortion Harmonics”. Each present is for one level of Distortion Harmonics, and has four captures - one for each of the levels of Sustain. A little chorus/delay/reverb for a more Rockman flavor but didn’t put a ton of time trying to emulate those. If anyone has great settings for those effects love to know what they are and will update the presets.
I’ve also put up my Bogner Atma on the “Crunch” setting with Gain full up on each of the three settings (60s/70s/80s) - it captures it quite well, and I’ll add some of the other settings over time.