From our news archives...

Artifacts ("bumps") at shot transitions

Regarding the Criterion Solaris, which I just viewed:
Picture quality is no doubt superb; however, why does there
seem to be a "splice" separating virtually every single shot
in the film?  The original Image Entertainment laserdisc was
also this way, and I've also noticed this phenomenon on
other films — most notably, for some reason, older
Japanese films of, say, early 60's Toho.

The net effect is that a "bump" occurs on virtually every
single "shot transition." In a film like Tarkovsky's, with
its dreamlike pace, this becomes extremely maddening and
— at least for this viewer — interferes with the
enjoyment of the film (I have not checked to see if the
Ruscico version has this flaw).

Clearly, these "splices" must be on the source print —
I'm just surprised that with all the attention paid to
digital restoration, that this is one flaw not corrected.
Has anyone else noticed?  By the way, I've also noticed a
trend in digitally restored films wherein, often, the image
will "freeze" for one frame just before the cut —
creating a similar "bump." I wonder if these phenomena are
related?  Perhaps this latter a failed attempt at digitally
correcting splices?  Films on DVD that spring to mind where
I've noticed: Rififi, Discreet Charm, Night of the Living
Dead.

[...] [People] I've watched films with usually don't notice
at first — arguably, it's on the subtle side.  But,
once I point it out, they simply can't stop noticing; it's a
pretty flagrant flaw once you know it's there.  

D.C.

We have also noticed these artifacts and do indeed find them highly distracting at times. The following are some letters we received in response to the above inquiry. Additional comments welcome.

Regarding your Jan 11 'jumping cuts' note — It does indeed
sound like an original cut negative problem.  With 35mm film
there is no need to 'zero cut' the negative, that is leave a
set number of frames on either side of the actual cut and
let the printer light actually make the cut.  Zero cutting
is used with 16mm optical blowups to 35mm as the space
between frames is so narrow that it's difficult to make a
clean cemented splice.  If there's a bump or a 'hard' cut
with 35mm — it may have been due to the registration pins
slightly shifting the film over a rough cement splice during
the contact printing procedure.  If the lab has misaligned
or not well maintained equipment, this could be a problem.

T.G.

You probably have readers far more technically-aware than
myself, but nevertheless:  Considering Criterion struck a
new print specifically for the Solaris transfer, I'm
guessing the printing machine they used was technologically
up-to-date and top-of-the-line.  Therefore, the problem
would likely be with the negative itself.

My guess would be that the negative was not cut particularly
well during the original neg conform way back when.  The
damage could have been done by a bad splicer blade, bad
adhesive, or just an incompetent neg cutter.  As a result,
the final frame of the outgoing shot & the first frame of
the incoming shot aren't going to line up perfectly, thus
producing the appearance of a "bump".

Offhand, it's difficult to say what could remedy the
problem.  But two possibilites.

1.  A.) Duplicate the second-to-last frame from the outgoing
shot, B.)  duplicate the second-frame-in on the incoming
shot, then C.) get rid of the offending frames in between.
I don't know for certain how this would look, but I doubt it
would be all that bad.

or,

2.  Once the film is scanned, attempt some type of digital
"re-alignment" of the frames being spliced together.

R.

Re.: Bumps
Has anyone determined what this is?  I have noticed this on
several DVDs, especially Hong Sang-soo's "Turning Gate,"
which I watched last night.  Every shot — rock steady and
then...  bump-cut-bump...  rock steady, etc.  What gives?

D.

The only plausible theory I can come up with is that it's
caused by an improperly cut negative.

The splices at each cut must be causing a bump as the film
goes through the machine used to duplicate the negative, or
through the telecine machine if the video transfer was made
directly from the negative.

Usually the video transfer isn't made from the negative.
It's made from a low contrast print or a low contrast
inter-negative (an inter-negative is a negative print).  But
none of the prints should have splices at each cut; only the
negative should have all these splices, so the original
fault must, I imagine, lie with the original negative and/or
the machinery used to make duplications from that negative.

D.

I notice a lot of these artifacts and find them maddeningly
distracting too.  If a film bounces at every cut, it is
clearly a fault not with a print, but with the negative or
the original transfer that was made from the negative, since
no other print would have all those splices.  This could be
very hard or impossible to fix.  A duplicated frame before
or after every cut does plausibly sound like an attempt to
fix this problem.

I've been annoyed by several of the Criterion transfers of
old color British films, like Olivier's Henry V and a couple
of the Powell/Pressburgers, where the color temperature
shifts up and down in a slow, pulsating pattern.  Anyone
else notice this or know what's caused it?

D.T.

I have both an early Pioneer dvd player (which I hope to
upgrade soon to something more state of the art), and a
Malata progressive scan player for other region discs; my
best and most "progressive" component is my Pioneer Elite
HDTV with film mode.

I'd be interested in hearing more about how a DVD player can
compensate for "bad edits," but I don't think the phenomena
described are related to player quality.  I've also
noted end of shot "freezes" on cable tv (a recent digital
restoration of a Laurel and Hardy short, and some costume
epic with Victor Mature I happened to come across during a
late night channel surf, for example).  The Criterion
'Solaris' doesn't SEEM to have digital freezes, but rather
raw splices in the film source.  'Night of the Living Dead,'
however, would be a perfect example of the digital freeze
I'm talking about.  Right around the end of the opening
credits, let your eye focus on any "moving" element (the
car, etc.); watch as it "freezes" just before the cut.  I'm
only surmising that these freezes were done to compensate
for raw splices (as found in the Criterion 'Solaris,'
'Discreet Charm,' etc.).  But it doesn't really seem to be a
viable solution: in either case, you get a visual "bump."

Coincidentally, I finally popped in the Ruscico 'Solaris'
last night to compare: the raw splices don't seem to be
there, or somehow compensated for so they're not as evident.
However, the image quality was neither as sharp or as
"stable" as the Criterion (I noted lots of pixilation on
camera pans, etc.).  Oh well.

D. C.

Subject: SPLICE BUMP ISSUE FROM JANUARY 11, 2003
Date: February 21, 2004 4:16:10 AM MST

First, thank you so much for your detailed, thoughtful, and
informative website.

As an film element inspector for Foto-Kem's Restoration
department, I wanted to add my two cents on the issue of
"bumps" during scene changes.  I offer my current insights
as a relative newcomer to the restoration field.

Almost all video transfers use timed "second generation"
elements (hopefully minted from the original negative): B&W
Fine Grain Masters, color low-contrast prints, or color
interpositives.  Whatever imperfections exist in the source
element will obviously transfer to the copy.

In the case of splices, many things can go wrong (and I've
seen a lot of them first hand) - poorly aligned splices,
thick splices, perf repair tape at the spice point making a
thick wad or "speedbump".  Almost all splices display some
level of pitch difference, or the relative alignment
difference between two adjacent perfs when spliced.  Most of
these are not obvious to the naked eye and require special
lupes that read in micon increments.  Therefore, most newly
spliced negative produces scene transitions that seem
smooth, at least to the typical moviegoer.

Older elements, however, have special issues that could
effect this (outside of the damage issues already
mentioned).  For example, film shrinks over time due to
moisture loss.  Sometimes, film can shrink to above a full
perf width its original size, causing the element to buckle
and pop off the printer wheels.  Often, slow-speed printing
or optical printing can produce an acceptable transfer
element.  But shrinkage also exacerbates the already
existent pitch differences between shots.  Old black and
white negative is especially vulnerable since it's thinner
(no dye layers) and often more brittle than color negative.
Also, some elements have been subject to re-cutting in the
original negative (or horrid idea).  When a splice is
re-done at the same point (a splice-over), a very thick
splice can result.

Telecine transfers, as far as I know, can do little or
nothing to correct this.  They have to work with what they
have, warts and all.  In fact, the more cutting-edge the
telecine machine, the more noticeable the imperfections.
High end digital transfers are often "too good" in picking
up every tiny detail.  I'm reminded of the first time I saw
High-Def TV displayed at Circuit City - I could make out the
tiny stitches in Barney the Dinosaur's costume.  Think of
how that kind of resolution can show poor transitions
printed into a timed fine grain.  Hopefully, every effort
was made to produce the best transfer element possible.

As for "freezes" in DVDs:

A company rep said that many DVDs often still- frame
"non-active" frames to save memory space on the disk.  If
the scene involves no movement (say, a background shot with
no actors), they may simply "hold" a representative frame
during the rest of the shot (usually a fraction of a second,
sometimes more).  This practice can produce some annoying
results to say the least, as many of your readers have
commented.

I'm assuming this happens in the authoring process, but I'm
not a DVD expert.

Troy Perry
Los Angeles, CA