I am trying to understand what could be an anomaly in my system.
96 kHz 32 bit F file in a montage.
Insert a random Steinberg plug in output … Master Rig for example and make a single token adjustment.
Then make a copy of that montage. Copy the plugin from the Output to the Clip effects in the copied montage. Delete the Output plugin so that there is only a clip effect.
We now have two montages: one with an effect in output the other with the same effect and setting in clip … otherwise the montages are identical.
Render them as 96 kHz 32 bit F files.
If you then compare the two renders (Analyze>Audio File Comparator) there are differences.
Why would this be the case? I did the same test in WL 9.5 and the renders are identical.
It’s not a practical issue as I’m only going to use clips or output but I’d like to understand it.
But seriously: I can’t explain the differences either, but I can confirm them. Binary (not audible) different audio files are created depending on whether the same effect with same preset is located in the MASTER, OUTPUT, TRACK or CLIP when rendering. Incidentally, there are no differences, if the same configuration is used several times in a row to render, i.e. results are reproducible.
I did the test you mention, with the MasterRig and a simple preset, but in my case, I get two identical files bit ti bit.
My montage is simple: just a clip.
Probably yours is more complex, with some gain adjustment somewhere.
Create a new, empty Audio Montage, Custom, Stereo 44100Hz. Insert an audio file (= just one clip). Save montage. Save as… 2 more montages, results in total in 3 bit-identically audio montages (checked the bit-identity). In first montage, insert the Steinberg VST MasterRig with preset “AM - Pop Rock” in the OUTPUT section. In the 2nd montage, do the same in the TRACK section and in 3rd montage insert it in the CLIP section. Now render all the 3 montages with identically settings into 3 audio files. Compare the 3 files in WaveLab or with an external app (e.g. Hex Editor or Hash value). Result: Every file is different from each other, e.g. CRC32: OUTPUT 75230E70, TRACK 77D3325E, CLIP C180D9F9.
I did some more tests and you are right. But it depends on the plugin or plugin preset. At first sight, only plugins with latency are concerned. I will have to delve deeper to try figuring it out.
I should add that the reason this came to my attention was: a while back I thought I could hear a difference … and in that case I was (surprisingly) ‘correct’. Here is the ‘difference’ in an old fashioned null test.
I initially put it down to the Voxengo Limiter plugin that used asymmetry in the process and didn’t have time to explore it further. But I also tried with a stock WL plug with same but not as noticeable results. At that point I thought it was my system having a bad digital day. But a similar test in 9.5 was OK (no difference).
As I said, it is possibly of no practical consequence … in that you are unlikely to have the same plugins on Track and Clip FX in the one montage. FWIW I had a personal preference (highly subjective) for the Track FX
Wow, I consider this a serious problem. Repeatability in mastering is fundamental.
I performed some tests using a stereo Sine Wave with different plug ins in 4 places: Clip, Track, Montage Output and Master Section. I then pulled them back into null tracks within the parent montage. I did not create separate montages. Rather I just moved the plug-in placement within the same montage.
Steinberg Stereo Tools. I raised the left channel 1dB. All four rendering permutations nulled against each other. However, against the parent file (FX in clip) I had problems. The unaffected Right channel nulled perfectly but the effected Left channel had residual at 28bits.
FabFilter L2. In all permutations the rendered files nulled against each other and the parent track (FX in clip).
So maybe this does point to particular plugins? But it was reported earlier that no such error with WL 9.5. Puzzling.
As long as a plugin is in a given location (clip, track, …), you get the same result every time you render. In other words, there is no randomness, what you monitor during your session is consistent with the final result.
That said, I agree this is unexpected, and I have yet to figure out what could be the cause.
@WYCA In the real world, there’s probably no ‘problem’. The render of a clip FX loaded montage renders perfectly … the same every time. As does the Track FX montage. As does a montage with given combination of Clip FX and Track FX. So what you set up to render is what you get.
I only posted this topic to try and understand what was going on in my system and PG is looking at that.
