Hi,
i have "discovered" a strange thing.
I have ripped a CD, compressed to FLAC, replaygained
I have done the same with MPC too.

Furious is, that replaygain-album-peak of FLAC is 0.977203, but replaygain-album-peak of MPC is 1.062957

It is very bizzare i mean....

This is external clipping. The original file has a peak value of 0.98. But when you encode it, the waveform is changed, possibly causing clipping, as you see.
When you apply the replaygain clipping prevention, it prevents this file from clipping. If internal clipping had occured, replaygain could not have done anything.

I'm not expert in replaygain... someone will correct me if I'm wrong.

The Program MP3Gain has the setting "Apply Max no-clip Gain for Album". I would really like to see that feature available for use with ReplayGain.

With foobar2000 there is a "Use peak info to scale down tracks that still clip after applying replaygain" checkbox located in Preferences->Playback.

edit: typo

True, I've used it, but it's not the same thing. It would probably solve the problem for the person who originally posted.

Yes. If using albumgain it might (you're saying does?) hurt the album dynamics, I am assuming this is the difference that you are referring to. I haven't tested for this, but the possibility did occur to me.

edit: clarification

Lossy encoders slightly reconstruct the original waveform of a clipped signal.

When you take a sine wave, clip it, encode it and decode it, you get something between
the rectangular of the clipped and the sine shape of the original signal.

So you must reduce the amplitude or reclip the signal again.

Non-clipped signal has not this problem. Peak amplitude increase is a property of
heavily clipped signals.

Effect on clipped signals:



Effect on non-clipped signals: