|
|
Post by viciousbliss on Nov 14, 2022 4:38:58 GMT -6
I've been going back and forth with Paul Third on Youtube over his ID14 vs Access Analog's Lynx video. If I understood him correctly, he was claiming that a pure ITB bounce would not be different from one interface to another. Many on forums make this claim too. When I first got the Lynx, I did those tests and the files did not null. A Youtuber named Streaky did a Hilo vs UMC22 test and put the wav files up. Definitely not identical in the least. The UMC22 does not have hardware inputs, I don't see how it wasn't a pure ITB bounce. Just now I bounced two mixes with the 404HD from two of my latest sessions. Did not null, the vocals were still coming through after doing the invert test in Audacity. Did not sound the same either. The Lynx bounce was much more cohesive and full. I did get files to null from a mastering session where I had committed the tracks with Satin and the hardware Fusion and then added three plugins to that file. Just bounced the same thing with the 404hd. The mastering session was 96k and the mix sessions were 44k with a lot of oversampling plugins. I'm not sure if the same rate had something to do with it. I'm 100% sure I didn't change anything since these are the latest things I opened.
Earlier this month I did a test of a Twin bounce with a 404HD bounce from the same mix session and they nulled.
The only UAD plugin I had open in the mastering session was the Massive Passive. In the mix sessions it was a lot of their FX. I'm not sure if the Aurora's clock comes into play with an Octo Satellite.
So, what's the verdict on bounces from different interfaces?
|
|
|
|
Post by svart on Nov 14, 2022 10:34:40 GMT -6
I've been going back and forth with Paul Third on Youtube over his ID14 vs Access Analog's Lynx video. If I understood him correctly, he was claiming that a pure ITB bounce would not be different from one interface to another. Many on forums make this claim too. When I first got the Lynx, I did those tests and the files did not null. A Youtuber named Streaky did a Hilo vs UMC22 test and put the wav files up. Definitely not identical in the least. The UMC22 does not have hardware inputs, I don't see how it wasn't a pure ITB bounce. Just now I bounced two mixes with the 404HD from two of my latest sessions. Did not null, the vocals were still coming through after doing the invert test in Audacity. Did not sound the same either. The Lynx bounce was much more cohesive and full. I did get files to null from a mastering session where I had committed the tracks with Satin and the hardware Fusion and then added three plugins to that file. Just bounced the same thing with the 404hd. The mastering session was 96k and the mix sessions were 44k with a lot of oversampling plugins. I'm not sure if the same rate had something to do with it. I'm 100% sure I didn't change anything since these are the latest things I opened. Earlier this month I did a test of a Twin bounce with a 404HD bounce from the same mix session and they nulled. The only UAD plugin I had open in the mastering session was the Massive Passive. In the mix sessions it was a lot of their FX. I'm not sure if the Aurora's clock comes into play with an Octo Satellite. So, what's the verdict on bounces from different interfaces? Not sure what you're asking here. ITB renders mean that the entire render is done in the DAW with zero external devices in either DSP or analog domains. You should be able to render a session even without an interface present in this case. So I'm not sure what you mean by "bounce" with an interface present. Besides, one very overlooked situation with null tests are the nonlinearity of some plugs in terms of modulation. They might not modulate the same way every render, so they'll never null when compared later. It's very common with delays and reverbs and one of the most heavily delayed and 'verbed tracks in a mix are the vocals, so it's possible vocals will remain standing out in a null test. Also, sometimes plugs can change their DPC latency if there's a lot going on in a mix or if there's background processes happening that can interrupt time-critical processing. Null test renders should always be performed without ANY plugins for all these reasons. |
|
|
|
Post by bchurch on Nov 14, 2022 10:47:51 GMT -6
Yeah, I am admittedly confused - perhaps at the wording of the question. What I'm reading is that one bounce was using a plug-in, and another was not. Which, obviously, is going to make a difference.
I suppose that conceivably using two different interfaces = two different clock sources. But a profound effect and difference? Likely not.
|
|
|
|
Post by recordingengineer on Nov 14, 2022 11:38:31 GMT -6
Indeed. I read with an open-mind, but the true situation and question is unclear, as a true traditional PT Bounce would not have anything to do with an interface.
Here’s a question I don’t know the answer to:
Today, is there any difference between a PT Native Bounce and a PT HD Bounce? I remember in the earlier years, there was.
|
|
|
|
Post by sirthought on Nov 14, 2022 13:31:36 GMT -6
I think what you guys are saying makes sense, but then again, it's something I don't know anything about...does your DAC effect the sound of an ITB bounce?
I'd say no, as the DAC is just impacting what you hear from your monitoring source.
I'd be curious to hear Streaky's explanation of what he thinks is happening between the two different interfaces. The guy certainly has experience with gear I don't.
|
|
|
|
Post by viciousbliss on Nov 14, 2022 14:15:13 GMT -6
I’m just saying I have the Lynx hooked up, bounce a file without using hardware and then hook up the 404hd and bounce again. The files don’t null when compared. Part of the vocals still play. Streaky does a DA test of the UMC202 and the Hilo and the wav files sound very different. I’m still waiting to see what he says was his method for rendering the files.
I did get bounces from a mastering session to null between the 404hd and the N. Those had Satin and Fusion printed and then UAD Massive, Unisum, and Oxford Limited added after. And they were at 96k as opposed to the mix bounces that didn’t null, which were at 44k.
There was also a lot of UAD on the mixes that didn’t null. I’m not sure if there is any DA involved with processing UAD.
|
|
|
|
Post by tkaitkai on Nov 14, 2022 14:18:11 GMT -6
Another consideration here whether you’re rendering online vs. offline. In Reaper, for example, there is sometimes a noticeable difference between online vs. offline renders. But this probably has less to do with the DAW itself and is more a function of how (as svart pointed out) the plugins used within the session are communicating with the DAW during online vs. offline renders. FWIW, I now opt for online renders when doing final revisions. I want to be 100% certain what I’ve been hearing is what’s being rendered. |
|
|
|
Post by svart on Nov 14, 2022 14:32:09 GMT -6
Another consideration here whether you’re rendering online vs. offline. In Reaper, for example, there is sometimes a noticeable difference between online vs. offline renders. But this probably has less to do with the DAW itself and is more a function of how (as svart pointed out) the plugins used within the session are communicating with the DAW during online vs. offline renders. FWIW, I now opt for online renders when doing final revisions. I want to be 100% certain what I’ve been hearing is what’s being rendered. I've heard this too, but it was a specific plugin that did not like full speed offline renders. Once I turned it down to 1x it didn't matter if I did online or offline, it sounded the same after that. Usually if I'm doing a bunch of renders of regions in a single project I'll do one full speed offline render to see how it sounds. If it's OK, I'll just continue for the rest of the regions. If not, I'll switch to either 1x offline, or online for that project. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 14, 2022 14:58:00 GMT -6
UMC22 is a another poc Behringer product that Uli was too cheap to license ASIO drivers for. Believe me the drivers and the usb chipset are the most expensive thing in the more expensive Behringer interfaces.
Since it doesn’t use ASIO, anything is possible under Windows. ASIO4ALL is not ASIO but merely a wrapper for DirectSound.
|
|
|
|
Post by viciousbliss on Nov 14, 2022 15:37:23 GMT -6
I’m pretty sure the 404hd uses ASIO. If I recall, it’s listed as UMCASIO in the interface selection. Going back to Access Analog, if you use their Lynx Aurora N inside the Analog Matrix plugin, it will give that Lynx sound on top of what your already getting.
If I wanted to hook the Twin or ID14 back up, I could do more comparisons. I did previously get separate Twin and 404hd bounces to null from the same 44k mix session. Satin’s manual said something about a different result on every bounce or whatever if I recall, but I don’t think that’s what’s going on here.
Here’s the Streaky video:
|
|
|
|
Post by craigmorris74 on Nov 14, 2022 16:04:19 GMT -6
The Streaky video compares D/A converters, which are not involved in a render at all.
To prove that I wasn’t crazy, I just did 3 renders, one while my UA2192 was clocking while attached to an RME PCI card, one with the a shitty (and the DA sounds TERRIBLE) Behringer USB interface, and finally one with the windows audio drivers. All three renders nulled to infinity.
If you’re using a plugin that purposely generates random responses, like some reverbs, delays and other effects, two renders with the same device will sound different. So if Satin is introducing random responses, that’s why files don’t null.
|
|
|
|
Post by viciousbliss on Nov 14, 2022 17:15:00 GMT -6
The Streaky video compares D/A converters, which are not involved in a render at all. To prove that I wasn’t crazy, I just did 3 renders, one while my UA2192 was clocking while attached to an RME PCI card, one with the a shitty (and the DA sounds TERRIBLE) Behringer USB interface, and finally one with the windows audio drivers. All three renders nulled to infinity. If you’re using a plugin that purposely generates random responses, like some reverbs, delays and other effects, two renders with the same device will sound different. So if Satin is introducing random responses, that’s why files don’t null. I’m just curious why his wav files sound different. The video doesn’t say how he rendered them. The differences I’m getting are too big to be from a random change in Satin noise. I can go and bounce again and see what happens. I’m pretty sure that I tested different Satin bounces years ago and found that they nulled. There may be something going on with the Lynx or how it setup the default routing. I may email them. |
|
|
|
Post by christopher on Nov 14, 2022 18:36:50 GMT -6
Vocals are usually centered so first thing that I think to look into is see if panning laws are staying true, or if device drivers cause a change there, or mid side or LCR is somehow part of the algorithm. There’s so much stuff under the hood in DAWs that won’t won’t show up on meters. Many DAWS like Reaper the meters don’t show down to -Inf, so there can be plenty of differences that people think are perfect nulls. The meters are totally blank, yet the noise floor is there somewhere
I just record the Monitor outputs onto a stereo 24/96 wav. I figure I’ll at least be recording exactly what I’m hearing and not something else.
In the end whatever hit there is has to survive so much.. phones, steaming.. cars.. TVs.. it’s always gonna be a compromise I guess
|
|
|
|
Post by viciousbliss on Nov 14, 2022 18:45:39 GMT -6
Well I bounced a duplicate Lynx file and it also didn’t null with the previous Lynx bounce. Same thing where parts of the vocals were leftover after inverting. I don’t think it’s just Satin since I used Satin on the music track too. I don’t know if UAD Lexicon or Capitol Chambers would cause variations like that.
I wonder if Streaky recorded the monitor outs back in or something like that because he definitely doesn’t mention that offline bounces should be identical no matter the interface.
|
|
|
|
Post by christopher on Nov 14, 2022 18:54:37 GMT -6
The plugins could be using MS. I sometimes experiment with Reverb and Delay make compressing on the side channel different, so it kind of expands out. But you’d think they’d be perfect math, but I bet Svart and others are right in that they use a randomize algorithm to create a better dispersion? So each pass is unique?
|
|
|
|
Post by viciousbliss on Nov 14, 2022 19:23:33 GMT -6
I wonder what you get back on Access Analog, because you’re definitely hearing the sound that I get from listening to the Lynx on headphones.
So, I guess it’s real value is in monitoring, converting hardware, and recording? That’s where all the conversion happens it seems.
I haven’t found any definite info on whether UAD usb satellites use conversion or not. I’m guessing they don’t. I could make duplicate bounces with the 404hd and see what happens.
|
|
|
|
Post by craigmorris74 on Nov 14, 2022 20:34:03 GMT -6
Satin, Capitol Chambers, and maybe UAD Lexicon (never used that so don’t know) will never render the same way twice. That’s part of their “analog charm”.
Streaky never says he bounced/rendered the file. He doesn’t explain how he generated the files, but he most likely ran the file out the DA of the Behringer, and recorded the result, then did the same thing with the Hilo. The DA converter of the PLAYBACK was captured.
For one final test, I bounced down a 36 track mix (had to turn off Reverberate and Pulsar Echorec, because they give random results) with my UA2192 as the DA, then completely disabled the audio in my computer from the device manager and rendered the project again. The two files nulled to infinity.
And yes, Reaper nulls to infinity.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 14, 2022 20:45:16 GMT -6
U-he Satin does things the right way and oversamples enough (384 khz) to have a bias tone unlike most other flavor of the month tape plugs where distortion increases when you over "bias" them. Satin also has tape hiss, asperity, and wow and flutter, all of which are random.
|
|
|
|
Post by Blackdawg on Nov 14, 2022 22:55:18 GMT -6
It sounds like you have a plugin or several that have some features taht make them never render the same. Start by disabling them one at a time.
Also you keep saying you're using access analog, isn't that the service that you use a plugin to tap into their analog hardware? In which case, that's why things aren't nulling. If you are using hardware of any kind it will not null. Because analog..
Streaky was comparing DA converters. So he played something out of the DA outputs of each device and recorded that. Then showed how they are different.
ITB rendering with no analog hardware and no weird plugins that are non linear all the time will always be rendered the same. There is NO rendering/conversion in an ITB mix. Ever. It all just 1's and 0's.
So what I would do to properly test this is just grab a stereo wav file of something. Slap a basic EQ on it. And render it out. Then do it again. And see if they null. I'd be money they do.
Then go back to your session and analyze what plugins you are using and how and start turning them off one by one. You'll find the ones that are causing your results to not null. It could be anything that is an analog emulation plugin. Which most now have a button to disable "analog" and then it'll probably null.
IF you are wanting to compare the DA of one interface to another. THen you need to route out of said interface and then back into another to be recorded. Do this for the other and use the same interface on the input and it'll show you the differences.
|
|
|
|
Post by lando on Nov 15, 2022 4:28:57 GMT -6
Well I bounced a duplicate Lynx file and it also didn’t null with the previous Lynx bounce. Same thing where parts of the vocals were leftover after inverting. I don’t think it’s just Satin since I used Satin on the music track too. I don’t know if UAD Lexicon or Capitol Chambers would cause variations like that. I wonder if Streaky recorded the monitor outs back in or something like that because he definitely doesn’t mention that offline bounces should be identical no matter the interface. It’s highly likely that all of the mentioned plugins have random elements. Reverbs in particular often has this. Likely this is the source of the files not matching. What cannot be a source is the audio interface, as it is not involved in an offline bounce. |
|
|
|
Post by svart on Nov 15, 2022 7:36:26 GMT -6
Vocals are usually centered so first thing that I think to look into is see if panning laws are staying true, or if device drivers cause a change there, or mid side or LCR is somehow part of the algorithm. There’s so much stuff under the hood in DAWs that won’t won’t show up on meters. Many DAWS like Reaper the meters don’t show down to -Inf, so there can be plenty of differences that people think are perfect nulls. The meters are totally blank, yet the noise floor is there somewhere I just record the Monitor outputs onto a stereo 24/96 wav. I figure I’ll at least be recording exactly what I’m hearing and not something else. In the end whatever hit there is has to survive so much.. phones, steaming.. cars.. TVs.. it’s always gonna be a compromise I guess Reaper does, just not on the meter graphic. The peak value in decimal above the graph will show down to -INF. |
|
