sorry
My bad assumption.
And yes, I've read the thread where you, Karl and Andrej discussed it and I couldn't make my mind on what you all found out or agreed on .
Understood.
What do you mean with "by default"?
Isn't it the "Output YUV data as Basic YCbCr not CCIR601"?
If so, should I set it to on or to off? I believe the default is off (unticked), isn't it?
Ok then. So if I want to compare the two of them (in fact along with NuEnc) I should enable the above Tmpgenc option so that Tmpgenc doesn't clamp the chroma/luma values, right?
Even because HC and NuEnc don't have such a feature themselves, right?
Will do so, for the testing purposes but what about in real life encodings?
Isn't that bad for the TV set? I mean not with small testing clips but with a lot of big >=2h00 movies?
You mean that you're not sure if Tmpg cuts or rescales the ranges?
BTW what's the difference?
I'm going so much off topic that I wonder if we shouldn't split the thread at this point
Sorry about that, guys.

__________________
Rui

As you say, the default is off, means "TV scale" and it's the setting to use.
(this can be discussed if you have somthing like a plasma screen)

That sounds correct.

What I told you is for real life encoding.

You missed something : I told you to use the option for luma clip in TV range for the encoder. So your TV can't be harmed.

No I am not and that is what Inc tries to find out but as yourself, I'm sure sure about his conclusion

Take something taht is between 0-255, you can cut the part at both ends to have something between 16-235 or you can shrink the values (by a mathematical operation) to tuen then 0-255 into 16-235. First is clipping, second is scaling.

Did some tests with the movies kwag made.

There are some differences:

Luma scale: TMPGenc uses a clipped scale of 16 - 235, HC uses 0 - 255.
Took the same frame (both B-frames) and put them online:

TMPGenc frame: http://hank315.dyndns.org/KVCD_TMPGenc.jpg
HC frame: http://hank315.dyndns.org/KVCD_HC.jpg

Bitrate: Bitrate distribution looks almost the same but what puzzles me is the difference in the quantisation values. All settings seem compatible, this would mean the TMPGenc encoding should be much better because of the lower Q values but I can't see that big difference.

Please look at: http://hank315.dyndns.org/KVCD_bitrate.jpg

I never used TMGenc so don't know the settings but it seems it uses some kind of filtering so the content can be compressed more easily.

grtx. hank

There something else that puzzle me in all these snapshots : in both cases (audioslave post and now in yours, hank) the picture form tmpgnec does not have the same A/R (or resolution, I don't know) than the one form HC !
It's very obvious when you open it in two tab under mozilla and change from one to the other.

Is that normal ?

Just did a clipping cut somewhere in the black bars (vertical direction), just checked it but the originals are really equal.

EDIT: only equal in size

@Hank
Is it possible for you to add a "Luma Selector" in the new HC version? From my experience a luma scale of 16 - 235 looks much better than 0 - 255 when viewed on a TV screen.

@Dialhot
About the aspect ratio: The screenshots I posted looks, and are, identical in regards to the AR?! But, the ones that Hank posted were different...

__________________
AudioSlave

@Hank,

Feature request: Could you add 3:2 "Pulldown" on a future release
It would be nice not to have to run the .m2v through pulldown.exe after encoding

Also, I second Rui's comments about avalon. He was kicked out of this site for stealing other people's work and claiming it was his work, and it's all well documented in this forum.
Just wanted to let you know

-kwag

Ok Phil,
I'm still digesting what you posted and I am having a hard time to understand it. I will post details about it.
But what puzzles me the most is that you're actually very strong when you say tmpg clip is darker than HC clip.
And that both audioslaves's pics are different in A/R.
I can't understand how you're getting such differences.
For me it is very clear that tmpg is darker than HC.
It may or not be related to luma/chroma different scales but tmpg looks darker.
I don't know if it's my eyes or if it has something to do with my screen but that's what I got.
On the matter of audioslaves pictures, whichever way I open them and compare them I find the A/R to be the same/identical.
So I'm feeling very bad if it's only me getting this result.
Or is there anybody else in here having the same results?
I know that we're comparing oranges with apples but this is really what I got and I am having a hard time to tell my eyes and my brain that it is not.
I will get into details further on today.
I'm in a hurry to get to work
BTW it seems to be useless to post any comparisons between encoders until I get them to work in exact same conditions.
So I'll only post my results after I understand all that is going on here.
TIA Phil
Cheers

__________________
Rui

Okay.

I am talking about the SNAPSHOTS (png files). I did not DL the clip (m2v files).

Not the snapshots from audioslave .

I just checked with Audioslave's one and they are identical (in size) "today" . I think I had a problem on my PC with an "autozoom" extension for Mozilla that did not show the picture at 100% size. So drop this point. But the tmpgenc snapshot is still lighter !

As HC continue to hangs on the second open of the same avs on my PC, I can't do anything but wait for the next release.

Hank ?

Another bad call from me
Because I was not trying to refer to the clips (m2v's) I was trying to refer to audioslave's png pictures.
Using my home PC and my work PC I see Tmpg's picture darker than HC's picture .
Will someone else please report about these pictures, please?
Anyone, please.

Good, at least we made some progress on the AR we now see the same thing and have the same opinion about it.
As to the picture's darker/lighter I already expressed myself above.

That's odd.
Anyone else with same issue?
I can open as many avs as I want
I guess you would be using a simple Mpeg2Source() and not many options right?
That is, you've tried HC on more than one script, right?
Any clues on this one Hank?

__________________
Rui

Listen, I took the left part of the tmpgenc snapshot and put beside it the right part of the one from HC. If you continue to see the left darker than the right, then we have a english problem

What can I say
I don't think that I can clearly say something about that picture.
If I take a look at the ground's color (orange dust) I would say that it looks darker on the left side than on the right side...
But it's a tough call because it could be just like that in the source
My terminology assumptions:
Darker: meaning that colors tend to get closer to black than to white. Usually harder to see details.
Lighter: meaning that colors tend to get closer to white than to black. Usually easier to see details. As if a spot light was illuminating the subject.
Are these assumptions correct?
I am almost ready to accept that I am wrong here
Will somebody else please say what they think about this color dark/light issue?
Cheers

__________________
Rui

Lol. Look at the original pictures : the sand is darker on the right than on the left for both picture ! The light is not uniform in all the picture
Please rui, do open the two pictures in 2 tab and switch from one to the other, tmpgenc is definitely lighter !

We never talked about the source. The point here is just "one is lighter than the other" with no assumption on which one is correct and which is not.

Yes.

The logic of the file buttons isn't very good ATM, this will be solved in the next version, also Avisynth errors will be reported.
But starting the encoder again and opening an AVS file should work.
I hope to finish the new GUI in about a week, most of the errors are already solved and the preview section is almost finished.

About all the other questions, everything is possible if I could find the time...

Now I'm going to code all evening

grtx. hank

It is not the case on a win2000 PC and a script using avisource. The GUI hangs for ever insteed on the second load of the avs. You have to reboot to restore the situation.

That's why I said I wasn't sure when you posted your pict with two halfs of the 2 original pictues
That's mostly what I've been doing, I assure you.
But!...this last time that I opened both pics side by side, I thought about the definition of darker and lighter.
Darker with stronger colors and less defined details.
Lighter with weaker colors and more defined details.
That's when I saw that I have always been looking at the top of the picture, especially to the sky.
It appears to me that the sky is lighter on HC's pic than on tmpg's.
But then I took a look at the trees, wagon, floor and I saw what you've been trying to tell me all this time.
On tmpg's pic I can clearly see the top of the trees on the left side, and the details of the left side wagon back door.
I cannot do the same on HC's pic.
And I also saw that the floor's orange color is weaker on Tmpg's pic.
So now I'm starting to think that you were right all the time.
But what about the sky???
Why is it clearer on HC's pic?
Maybe this isn't really a darker/lighter issue.
Maybe this has something to do with color saturation or contrast or whatever.
But now if you tell me that Tmpg's pic is lighter, I will nod 'yes' even if I'm not 100% convinced that it has to do with dark/light
So sorry about that, old fellow
What is absolutely true, although I think it would be wise to proove which one is closer to the source not just the pics the whole movie clips.
But I'm sure you'll guess that it is tmpg's the closer one.

@Hank
Sorry for going so much off topic here buddy.
Now that we're getting very close to the bottom of it, and if we proove that HC encodings are darker than the sources, I wonder if you could add to the very bottom of your to-do list a feature to cut (or scale or whatever) the luma/chroma levels to TV set acceptable ranges?
One other thing: I've noticed that when I load HC encodes in bitrate viewer it always reports "Nom. bitrate: 9800000 Bit/Sec" even if I set the max bitrate for 5.000.000 or 8.000.000.
That hasn't represented any problems when muxing or anything (why should it, right?).
But just in case, could you take a look at it one of these days?
And maybe it would be nice if HC could change the framrate on the encoded clip. I don't need it but others might find it cool.
I know
We should start a thread for the wish list
Now how about that?
Cheers

__________________
Rui

Hey Rui,

My samples were just to show HC's good motion estimation
Not to cause a color space havoc I'll answer that
You can use DVDPatcher to "Brand" the mpeg file to whatever you want

-kwag

I wanted to tell in my last post than light/dark was not the good word. This is more about vivid or not. "Vivid" is a combination of luma and saturation. So you're right

When I look at the pictures I really don"t know. Mainly because I found the (darker) HC picture more pleasant to watch so I have the whish this is the correct picture

@Karl
The samples reveals others things, it is not void to discuss these point also.

Yes, it's fine
But that was not my intention

-kwag

Tried to replicate this error but my PC runs all tests OK, running XP, SP1. Never got it so far that I had to reboot...
Think (hope) the next version will solve all this.

About the 9800000 bits issue: this has a relation with the size of the VBV buffer which I set at the maximum. And yes, you could run into trouble if you want to mux it with multiple high bitrate audiostreams but AFAIK the muxer will look at the actual bitrate and not this bitrate value.
It's just a maximum value, the upper bound of the coded data which is fed into the VBV buffer.
But please, correct me if I'm wrong on this, it can easily be changed.

A comparison is difficult because of the different Luma scales.
I found the HC pics also more pleasant to watch, not because it's my own encoder, still try to be objective .
But, I watch them on my PC screen, on a TV screen it might be completely different.
With CCE you can also choose a 16-235 scale, but it will not produce the same "foggy" output as TMPGenc, still think it's some kind of filtering...
But again, don't know that much of TMPGenc, it's just a first visual impression (on a PC-screen )