Jump to content

maxmercy

Moderators
 View Profile  See their activity

  • Posts

    2,092

  • Joined

  • Last visited

  • Days Won

    150

 Content Type 

Everything posted by maxmercy

  1. maxmercy
    FM, The soundworks site is very good, nicely done videos. What or where is the best way to find out who the main sound designer and mixer are for a particular film? If the end-title credits can be sometimes wrong, how can we find out who to properly credit? I'm sure you understand we are doing this in our spare time, and errors can be a part of this thread, as much as we try to minimize them. I appreciate you stopping by, because while we can speculate as to why a film graphs a certain way, you may have the direct means to explain why it is so. JSS
  2. maxmercy
    Ricci, With a 2Hz bin width, we halve the FFT time to 1/2 sec, and halve the resolution. The rise time is also halved, and the waterfall/spectrum graph is updated every ~30msec with a 93% overlap. With a 4Hz bin width, we have a 1/4 sec FFT, but only 40 entire bins for the entire LF bandwidth! Even though the graph is updating every ~16msec, it looks like a 1/3 octave spectrum analyzer..... I may start a poll on this before I graph any more films. I remember debating this a while back, but seeing a few films really get shortchanged recently that rely on mainly transient effects may move me to using a 2Hz bin width. I think 4Hz may be too much, unless combined with the 1Hz info, but then we are looking at two separate graphing sessions, as a film runs in real-time. If we are able to get input from several different 'measures', we may be able to do multiple passes on a single film to see what is truly there without being selective in the time or frequency domain.....I'll work on those tutorials this weekend. I'm sure we can all agree that scene graphs should remain hi-res though, correct? I think so. The value of scene graphs is knowing the freq range of an effect, with level being secondary. Thanks for the film suggestions, I will add them to the queue..... I do agree on doing a comparison on different versions of a film, with the stuff we have found so far. I have the M&C BluRay disc. JSS
  3. maxmercy
    I account for Dialnorm because it is something 'added' in for the home environment, and something that THX receivers will adjust for as well. If I left Dialnorm attenuation in, BlackHawk Down would have been 7dB low on the DD recording! I used PCM, which is curiously 7dB higher than the DD track...... Transformers 1 BluRay has -4dB Dialnorm, ROTF BluRays, one has -4dB Dialnorm, one doesn't. When you adjust the one that does up 4dB, the graphs are almost identical. I was posting the above, so I didn't see your comments earlier. What do you think of the FFT bin size? I think all scene caps should be hires, but the PvA should probably be changed to better show true levels, since Level, Extension, and Dynamics are taken directly from that graph. Do you think 2 separate bin sizes (wide bin for Level/Dynamics, narrow bin for Extension), or just a compromise in between bin? Also, I still cannot explain the differences you and I see in ALVH. It's actually pretty weird. JSS
  4. maxmercy
    OK, Here are some comparisons of three separate FFT settings. The settings are: 1. 1Hz resolution, but 1 second integration time 2. 2Hz resolution, but 1/2 second integration time 3. 4Hz resolution, but 1/4 second integration time All three have pluses and minuses. For example, in the 1Hz FFT, we can see the individual frequency makeup of an effect with great resolution in frequency, but short-lived effects, like gunshots, get 'averaged out' because the FFT looks at a second of data at a time, adding 63 miliseconds to each and looking at a one second interval again. So if an effect lasts more than a half second, it shows up as very strong. If an effect is very short, even if narrowband, the FFT will show it as lower in dB, because it got 'averaged' with the rest of the stuff going on that second. The lower resolution FFTs take less time to integrate, but lose resolution. So gunshots are shown at a more accurate level, but resolution is lost. Welcome to the FFT as an analysis tool. As your integration time drops, your resolution drops, but you can now can accurately graph the level of broadband, fast effects with more precision. I chose 3 samples. 1. Danley's Fireworks recording - This is 100% wideband, fast effects. You can see that the shape of the curve does not change much, even though the level definitely rises. 2. Disney's BluRay trailer - This is a mix of mainly fast, wideband effects, with a little narrowband effects too. Some explosions were given a broadband subsonic 'wallop' below 11Hz, and above 140Hz, which 'lingered' just a little longer with each explosion, so they show up VERY well on the 1Hz FFT, but not as well on the 4Hz FFT. 3. Dolby's Spheres Trailer - This trailer has a few booms, but two long duration effects with the strongest signal at 22Hz. , as well as a fast, LF downward sweep from ~50 to ~20Hz, varying in intensity as it goes. You can clearly see the single 22Hz contribution in the 1Hz sweep, but the fast sweep only spends a few miliseconds at each Hz, so the 1Hz, 1 second FFT doesn't have time to register it as very loud. Note that the sweep is recorded at a higher level than the sustained 22Hz tone, but the 1Hz FFT simply cannot do it justice. The shorter FFTs show poorer resolution, but definitely show the level of the sweep as larger, as the 'rise time' for each bin is shorter, and the sweep spends more time in each bin, beacuse each FFT bin is wider in the 2Hz and 4Hz FFTs. This may explain why films like the Incredible Hulk, Thor, WotW and ROTF have gotten such great scores with the current FFT scheme (1Hz, 1sec). If the effects are longer lasting, they will have more level with such an FFT. Star Trek, with it's tremendous warp scenes, got shortchanged. The warps are broadband effects, and would not show up very strongly in a peak graph unless a wider bin size was used. I propose we look at both 1Hz and 4Hz bin widths, or compromise with a 2Hz bin-width for PvA graphs. What do you guys think? I hope I explained this correctly.... JSS
  5. maxmercy
    Bosso, Thanks! I'll get them indexed soon. Did Thor clip your input? Also, ALVH may have gotten somewhat short-changed by the Hann window. I'll re-run when I get time with Nuttall. But I am still concerned that it looked like the levels you were getting were on par with Thor, but I got just 'meh' levels... Infrasonic, I'll get a tutorials up this weekend so ppl can help with screencaps and Peak/Avg. Later tonight I will put up a comparo of wide v narrow bin peak/avg for different THX and Dolby trailers for input. If enough ppl can contribute, we may be able to do both narrow and wide bin analysis so that short-lived effects do not get relatively shortchanged as they currently do by the peak/avg graph.... Ideally, we would use a wide-bin FFT for Level and Dynamics, and a narrow-band FFT for Extension and for better resolution. JSS
  6. maxmercy
    Oh jeez, guys. At some point over the last month, my SpecLab settings went from Nuttall back to Hann. This may help to explain why some of the latest crop of films graphed may be underwhelming. I will run a few of them with Nuttall windowing (it has a quicker rise time, so picks up brief effects better), to see if this is at fault, and to find out when the mistake happened. I hope not too many of the graphs will be changed.... Damn. But, this does open up a VERY good debate. With the current scheme, I can track resolution in frequency well (bin width of 1Hz). But Amplitude resolution suffers to a certain extent, because as the bin width decreases in size, the length that the program looks at the signal increases, 'masking' very quick effects that can be powerful, like the gunshot in ALVH, or the Gatling cannon shots into the bricks in The Dark Knight. By changing my bin width to 5Hz, quicker events can more easily be graphed, with a better picture of overall amplitude. Or, I could run both for each film, but that means running less films.... What do you guys think? I'll post up a comparison of a Dolby Trailer with each later today so you can see the difference. JSS
  7. maxmercy
    Very nice! I make a spreadsheet for every film, and have several other spreadsheets for all the dBHz data, a sheet just for graph comparisons, and one just for star ratings. My process is this: I graph the film using SpecLab (with Nutall windowing, thanks bosso!), using my BluRay player, connected to my AVR with HDMI, and I set the crossover points of all LCR/S speakers as high as I can (250Hz in my case). I also set the LFE low pass at 250Hz. I then have found settings so that a known reference track (soho54's Audio Test DVD's LFE tones) can be played back and analyzed for proper level and for any roll off in the signal chain, and to set accurate levels in SpecLab. If roll off is found, it is addressed by running a very slow sweep with Room EQ Wizard throughout the 0-160Hz range. This is then captured and analyed with DataThief, a shareware program. In order to capture data well, I have made my Speclab background white, peak and average traces black, and grid very light grey so DataThief can accurately graph the trace and export it to a text file with 0.5Hz resolution. Then I find the roll off point, and generate polynomial correction functions so that any further graphed material will be corrected properly after export to Excel. Once the correction functions are defined, you simply run the film while running SpecLab, then capture the screen, use DataThief to get the data into a text file, and import the text into Excel, make the proper roll off correction using simple cut and paste in the Excel file, and graph and analyze and save the data. That's the recipe. I can do the analysis for a single film in 15-20mins, and for multiple films in less time per film, hence me posting several at once. I have Speed Racer and SuckerPunch screen capped (both are tremendous visual films, but are very poor in level), but I am waiting for a few more so I can use time more efficiently. JSS PS - The reason some Avg tracings have an upward turning tail without the same in the peak trace is due to over-correction. Many of the lower level films will have such low <10Hz content that for my SpecLab settings (+5dB to -70dB), the graph is running flat for the bottom 2-5Hz, hence the upward tail due to correction applied to the 'flat spot' in the screencaps, vs a sharply downward sloped line. -10dB on my screencapped speclab graph correlates to 0dBFS, or 115dB. The reason I leave 15dB of headroom is that, for any reason, a 0dBFS signal would be sent in phase to every channel in a 7.1 system, I could track it. It would be over 125dB. No film has come close to even breaking the 115dB barrier at any one frequency, except for Attack of the Clones on DVD. When Dialnorm is accounted for, there is a 116dB peak in the low 40's, Hz-wise, when the Naboo ship flys by in the opening scene. The BluRay is lower in level by almost 6dB, IIRC.
  8. maxmercy
    Done. JSS
  9. maxmercy
    Will add them to queue. Should Pulse be included? JSS
  10. maxmercy
    Films coming up to be measured: Star Wars Saga BluRay, maybe comparos w/ DVDs LOTR Extended Editions BluRay Flight of the Phoenix How to Train Your Dragon Prometheus Speed Racer SuckerPunch Alien vs Predator The Dark Knight Rises Any other requests? Matrix Trilogy? After this, I am just gonna do new releases. I believe that the databass has a good base of data for comparison of new films. JSS
  11. maxmercy
    Bosso, Can you tell me more about your DVD? I got mine from Redbox, and my BD from Netflix. Both appear to be Rental copies without many special features. Here's the data: Abraham Lincoln: Vampire Hunter DVD Level - 1 Star (259.43dBHz) Extension - 4 Stars (15Hz) The BluRay ALVH was also double checked, and it does extend to 14hz on Avg graph, so it was changed as well. Dynamics - 4 Stars (26.6dB) Execution - 4 Stars A bit more level in this mix, otherwise quite the same. Overall - 3.25 Stars, same as the BluRay. Here's a comparo (attached): Looks like a 3-4dB level difference overall. Very similar otherwise. JSS
  12. maxmercy
    Max, It is because of how Level is calculated. The Level for a film is calculated on this thread by adding both the peak and average areas under the graph. This gives weight to the average level while giving a little more importance to peak levels, and singular but powerful effects. So, if Thor with dialnorm is 4dB low in Peak, and 4dB low in average, when added, they add up to 8dB total difference. Running the data on ALVH DVD.....will post soon. There is a difference. JSS
  13. maxmercy
    I'll be picking up ALVH today on DVD from Redbox to see the difference.... JSS
  14. maxmercy
    FilmMixer, Thanks for chiming in. I base most of what I know about Dialnorm off of this article: http://www.hometheaterhifi.com/volume_7_2/feature-article-dialog-normalization-6-2000.html So my knowledge is obviously limited going by the date of he article... I account for Dialnorm in all my charts, so every film is on a level playing field. I was just wondering if some folks have gotten a less than impressive presentation due to Dialnorm attenuation that people were not aware of, as I first did when I compared the Revenge of the Fallen regular vs big screen editions on BluRay. A -4dB attenuation overall is pretty significant when measured in dBHz over 4 octaves... My main question is why the obvious differences between mixes? It seems odd that one format would get a different mix over another....I remember you saying that most films have the theatrical mix on the disc. Is there a trend developing otherwise? I guess I should have said trailers are measured with A-weighting and left off the SPL meter part, but the concept is the same, it is a filtered measurement to measure the frequencies we are most sensitive to in lieu of others.... I have noted on several films that they 'left 7-10dB on the table'. Maybe the remix was for broadcast TV? JSS
  15. maxmercy
    Close. Dialnorm is something that authors use. It is a parameter set with a knob that comes set by Dolby at -27. When an author sets Dialnorm, they are basically saying that "in this soundtrack, dialogue is 27dB below the loudest sound possible (0dBFS)". The Dialnorm knob goes from -1 to -31dB. Authors who want to maximize dynamic range will use the -31 value, giving them 31dB to play with above dialogue SPL. When Dialnorm is set to -27dB (like most Dolby Digital content used to be on DVD), the author is saying that dialogue is mixed 27dB softer than the maximum allowable sound (0dBFS). To keep dialogue level consistent from show to show, Dolby will TURN DOWN the entire soundtrack by 4dB. Of course there is a problem. If authors do not measure their material, they are at a loss as to where to set it, and since it came from Dolby set at -27, many authors would simply leave it there. The other thing that is counterintuitive is that as you 'turn the knob up' on Dialnorm, it turns the soundtrack down.....which could have led to some confusion early on. With Thor, the difference in level between the 2 tracks is around 10dBHz. Playing the track back at +4dB, there is still a level difference, which means that the tracks are not equal to begin with. The DVD track has a different signature, ESP in the upper bass. My guess is this means that the DVD track underwent remixing for the home, in which dynamic range was altered. After the remix, dialogue level was 'altered' and was not as soft compared to the loudest effects. The mixers did the right thing, and set the Dialnorm to -27dB. That means whether you play the BluRay or DVD, dialogue is at the same level, but in the DVD, effects are softer by about 4dB, for playback 'in the home'. But you look at the graphs and say 'bullshit, when you turn up the DVD by 4dB, the low bass is almost identical'. That is because Dialnorm is set by an A-weighted SPL measurement. A-weighting does not look at low bass, it ignores it. You can see instantly in the graphs that the Thor BluRay has a lot more upper bass content compared to the DVD, hence the difference in measurement with A-weighting, and the subsequent Dialnorm adjustment. It looks like the mixers did the right thing. I just wish that there would be an option to get either the theatrical or the home mix..... As an aside, ever wonder why trailers are so bass-oriented? Back in the day, a trailer was made for a movie called "Empire Strikes Back", the loudest trailer in history, IIRC. It was after this that trailers were limited in loudness, but that loudness is measured by - you guessed it - A weighted SPL meters. Since A weighting ignores bass, you get a very bass-laden trailer if the authors want it loud, to get your attention. JSS
  16. maxmercy
    OK, here's the data. Thor BluRay - Level 269.5dBHz (5 Star), Dynamics 27.8dB Thor DVD with Dialnorm - Level 259.2dBHz (1 Star), Dynamics 27.5dB Thor DVD run at +4dBRef to account for Dialnorm - Level 267.18 (4 Star), Dynamics 27.5dB They BluRay and DVD are3 closer to eachother when dialnorm is level-matched. But the BD has a different signature, most notably the stuff around 140Hz that simply does not exist on the DVD. I check to make sure if any Dialnorm is applied in any film prior to recording data, as it shows up on my receiver's status display when the film starts. As you can see, Dialnorm can make a BIG difference. Black Hawk Down's BluRay defaults to Dolby Digital, with a -7dB dialnorm, neutering the track. The uncompressed PCM has no attenuation, and that is what I graphed... Here's what I think: The BluRay has the theatrical mix. The DVD has the 'nearfield' or 'home' mix, with Dialnorm attenuation added on top if it. I wish we could just get the theatrical mix and be done with it... JSS
  17. maxmercy
    Thor DVD vs BluRay is a big difference. Right off the bat, everything is turned down 4dB due to Dialnorm. They are simply different mixes. Will post data later. JSS
  18. maxmercy
    Bosso, I really think you are onto something. The only thing I can't figure out is why? Why the level differences? I checked during playback, and no DRC was on (that I could see)......I double checked levels with known material. At first I was suspecting 'rental' copy differences, but it may be something else.....I just can't figure why these folks would be doing something like this, throwing out 2 separate mixes for no reason, unless the theatrical mix somehow finds its way into one or the other, BD or DVD..... The more this kind of stuff goes on, the more you see that mistakes (and sometimes careless ones at that) are often made by these 'professionals', whether due to pressure from above (directors, execs, bosses, etc) or otherwise. JSS I will get the DVD for ALVH next week to see what is going on. PS - Now I see why I thought the ROTF screencaps were a little less than lively.....grrrr.... At least it is not as bad as music, where levels are even more all over the place....
  19. maxmercy
    Green Lantern: Level - 4 Stars (110.48dB composite) Extension - 4 Stars (12Hz) Dynamics - 4 Stars (26.07dB) Execution - 3 Stars - Very repetitive bass throughout. Although the lantern constructs have nice slam and heft, they are the exact same sound. It gets old. Quick. Overall - 3.75 Stars Recommendation: A reluctant Rent. Fans of the comic book may like it. But this one is probably better left on the shelf. JSS
  20. maxmercy
    As you can see, TIH and WotW are more similar than different, hence their overall rankings of 4.75 and 5 Stars, respectively. JSS
  21. maxmercy
    http://www.avsforum.com/photopost/data/2386425/ALVHTitleBlock-1.JPG http://www.avsforum.com/photopost/data/2386425/ALVHPH.JPG http://www.avsforum.com/photopost/data/2386425/ALVHCh-3.JPG http://www.avsforum.com/photopost/data/2386425/ALVHCh10a.JPG http://www.avsforum.com/photopost/data/2386425/ALVHCh10b.JPG http://www.avsforum.com/photopost/data/2386425/ALVHCh10c.JPG http://www.avsforum.com/photopost/data/2386425/ALVHCh22a.JPG http://www.avsforum.com/photopost/data/2386425/ALVHCh22b.JPG Abraham Lincoln:Vampire Hunter: Level - 3 Stars (109.88dB composite) Extension - 5 Stars (1Hz) Dynamics - 5 Stars (28.92dB) Execution - 5 Stars (by poll) Overall Rating - 4.5 Stars Recommendation: Buy (by poll) JSS
  22. maxmercy
    Edit 4-8-14: Updated measurements and PvA below. Wrath of the Titans: Level - 3 Stars (109.89dBHz) Extension - 5 Stars (1Hz) Dynamics - 4 Stars (26.9dB) Execution - 3 Stars - They left almost 10dB on the table, and dynamics took a hit, first 3 star movie in that regard. Either this is a 'home version' mix, or just too much of the same bass. I liked the film, and Ares hammer is very well done, but I did notice it was missing something. While ULF was there, ULF is not the whole story. This Star Rating is debatable. Overall - 3.75 Stars Recommendation: Rent. This film is good, but I don't know how many times I can see it enough to own it. JSS
  23. maxmercy
    Problem with Internet connection, wouldn't upload attachments, so multiple posts. JSS
  24. maxmercy
    Post Deleted due to Upload Problems. JSS
  25. maxmercy
    Post Deleted due to Upload errors. JSS
×
×
  • Create New...