Frame counter on sequences that don't start at frame 1

Questions and postings pertaining to the usage of ImageMagick regardless of the interface. This includes the command-line utilities, as well as the C and C++ APIs. Usage questions are like "How do I use ImageMagick to create drop shadows?".
Post Reply
ProfessorJerk

Frame counter on sequences that don't start at frame 1

Post by ProfessorJerk »

Hey there,

Trying to add a frame counter to a sequence of image files. It seems I can use -annotate %[scene] for this purpose for most sequences.
Unfortunately, the sequence also needs to be offset and when I add -scene the annotation gets screwed up.

Our architecture requirements impose the limitation that I probably need to do this in a single call (unless there is a way to modify the files in place with subsequent calls)

So this works because the resulting frames start at 1...
convert infile-%04d.tif[1-3] -annotate 0 "frame# %[scene]" outfile-%04d.tif

This fails because I try and offset the frame range...
convert infile-%04d.tif[1-3] -annotate 0 "frame# %[scene]" -scene 4 outfile-%04d.tif

Any suggestions?
User avatar
fmw42
Posts: 25562
Joined: 2007-07-02T17:14:51-07:00
Authentication code: 1152
Location: Sunnyvale, California, USA

Re: Frame counter on sequences that don't start at frame 1

Post by fmw42 »

Make a loop. For each image, find its scene number, offset it as new variable, then use that variable in -annotate.
ProfessorJerk

Re: Frame counter on sequences that don't start at frame 1

Post by ProfessorJerk »

ok so the best that I can tell, it looks like -scene destroys some per frame data.

I've tried to handle this by renumbering the images before I have to process them, so that I don't need to use the -scene option to do this.

convert infile-%04d.tif[4-8] -annotate 0 "frame# %[scene]" outfile-%04d.tif

This only seems to work on *some* images. That is, occasionally I have a sequence of image files that run from [4-8] as input, and the resulting output of the above statement is the creation of files with the same numbering [4-8]. *Other* image sequences that are the same format and also run from [4-8] result in output where the frame numbering is renumbered [0-4].

Why do some files get re-numbered and others do not? How can I tell if a sequence would be re-numbered or not?
User avatar
fmw42
Posts: 25562
Joined: 2007-07-02T17:14:51-07:00
Authentication code: 1152
Location: Sunnyvale, California, USA

Re: Frame counter on sequences that don't start at frame 1

Post by fmw42 »

I would always expect the former behavior that the annotation and numbering of the outputs would always correspond to the frame number of the input.


If I do:


convert rose: rose: rose: rose: rose: rose: rose: rose: rose.tif
convert rose.tif[3-6] -fill white -pointsize 24 -gravity center -annotate 0 "%[scene]" +adjoin rose3-6_%d.tif

I get images:

rose3-6_3.tif
rose3-6_4.tif
rose3-6_5.tif
rose3-6_6.tif

each has the numbers 3 through 6 correspondingly in the center of the image.

Is it possible that your input tif has more frames than you think.

Try posting a link to one of your input images for which you do not get this behavior.
User avatar
anthony
Posts: 8883
Joined: 2004-05-31T19:27:03-07:00
Authentication code: 8675308
Location: Brisbane, Australia

Re: Frame counter on sequences that don't start at frame 1

Post by anthony »

See my answer in the bug report...
viewtopic.php?f=3&t=16829#p62265
Anthony Thyssen -- Webmaster for ImageMagick Example Pages
https://imagemagick.org/Usage/
ProfessorJerk

Re: Frame counter on sequences that don't start at frame 1

Post by ProfessorJerk »

You can reproduce my current problem by doing the following:

1. Take any sequence of images that doesn't start at frame 0. (I'm using a test.0004.tif thru test.0008.tif )

2. Run imageMagick on them, the type of processing doesn't seem to matter. Here I'm changing the size of the image:

.\ImageMagick-6.5.9-Q16\convert ./Src/test.%04d.tif[4-8] +adjoin -gravity center -background rgb(0,0,0) -extent 960x720 ./Step1/test.%04d.tif

So far so good...we get ./Step1/test.0004.tif thru test.0008.tif. The frame range has not been changed.

3. Run imageMagick on the newly created frames. I'm running the same process, but I have tried others:

.\ImageMagick-6.5.9-Q16\convert ./Step1/test.%04d.tif[4-8] +adjoin -gravity center -background rgb(0,0,0) -extent 960x720 ./Step2/test.%04d.tif

Now the resulting frames are numbered [0-4]!

That is the issue I would like to understand. Previously I was adding -scene to my commands to ensure that it always started at the correct frame. But when I use that command the resulting %[scene] are incorrect, so that's not really an option to resolve this.

Thanks
User avatar
fmw42
Posts: 25562
Joined: 2007-07-02T17:14:51-07:00
Authentication code: 1152
Location: Sunnyvale, California, USA

Re: Frame counter on sequences that don't start at frame 1

Post by fmw42 »

1. Take any sequence of images that doesn't start at frame 0. (I'm using a test.0004.tif thru test.0008.tif )
Please clarify. These are separate images not separate frames. Each image could have its own set of frames or could be just one frame.

For example:

convert rose: rose: rose: rose: rose: rose: rose: rose: rose8.tif
identify rose8.tif
rose8.tif[0] TIFF 70x46 70x46+0+0 8-bit DirectClass 80KB 0.000u 0:00.019
rose8.tif[1] TIFF 70x46 70x46+0+0 8-bit DirectClass 80KB 0.000u 0:00.010
rose8.tif[2] TIFF 70x46 70x46+0+0 8-bit DirectClass 80KB 0.000u 0:00.010
rose8.tif[3] TIFF 70x46 70x46+0+0 8-bit DirectClass 80KB 0.000u 0:00.010
rose8.tif[4] TIFF 70x46 70x46+0+0 8-bit DirectClass 80KB 0.000u 0:00.010
rose8.tif[5] TIFF 70x46 70x46+0+0 8-bit DirectClass 80KB 0.000u 0:00.010
rose8.tif[6] TIFF 70x46 70x46+0+0 8-bit DirectClass 80KB 0.000u 0:00.010
rose8.tif[7] TIFF 70x46 70x46+0+0 8-bit DirectClass 80KB 0.000u 0:00.010
convert rose8.tif[3-6] -gravity center -background black -extent 100x100 rose8b_%d.tif

produces

rose8b_3.tif
rose8b_4.tif
rose8b_5.tif
rose8b_6.tif

What is wrong with that?

If you are trying to process multiple images to multiple outputs (via wildcards), you cannot do that with convert. You can use mogrify to process a directory of images to one output per image in the directory.
ProfessorJerk

Re: Frame counter on sequences that don't start at frame 1

Post by ProfessorJerk »

Thanks for the reply.
Please clarify. These are separate images not separate frames. Each image could have its own set of frames or could be just one frame.
My tests are separate images and separate frames. That is, 1 frame per image. The confusion here may be that I am using .tif as an example. This effect is reproducible with other image formats as well (png, jpg, bmp)

What is wrong with that?
Seemingly nothing. But that's not what I'm doing.
If you are trying to process multiple images to multiple outputs (via wildcards), you cannot do that with convert.
Isn't that what +adjoin does? I guess the part that confuses me is that it does work, provided I don't use images that were previously processed by ImageMagick. Using convert to create multiple images to multiple outputs works once, any attempt to use these generated files produces output that is renumbered.

I'll take your suggestion and look into mogrify, see if I can get it to do this.
User avatar
fmw42
Posts: 25562
Joined: 2007-07-02T17:14:51-07:00
Authentication code: 1152
Location: Sunnyvale, California, USA

Re: Frame counter on sequences that don't start at frame 1

Post by fmw42 »

Perhaps I misunderstand about your filenames test_%04d.tif[4-8]? Are these wild card sequences of one multiple images with multiple frames or is that just one image with multiple frames?

If the former, multiple images with multiple frames, then you are trying to get outputs from each image and separate the frames?

+adjoin only separates the frames of a given image into individual images rather than writing it as a single image with multiple frames.

I guess I am still confused about what you are trying to do. So until I understand better, perhaps my suggestion of mogrify may be premature.

I was not aware that one could do in convert the following:

convert image1 image2 image3... some processing result1 result2 result3...

Whether they had multiple frames or not.

For examples:

convert rose: rose: rose: -flip rose1.png rose2.png rose3.png

will not process at all. One gets error messages. You can use wild cards for input, so long as you have only one output (with multiple frames). see http://www.imagemagick.org/script/comma ... .php#input. But you cannot process multiple images to multiple outputs.

Using wildcards for input is the same as having multiple input images.

So can you clarify further with a simpler example?
Last edited by fmw42 on 2010-08-12T14:50:08-07:00, edited 1 time in total.
ProfessorJerk

Re: Frame counter on sequences that don't start at frame 1

Post by ProfessorJerk »

I have the following files...

Code: Select all

C:/Src/test.0004.tif
C:/Src/test.0005.tif
C:/Src/test.0006.tif
C:/Src/test.0007.tif
C:/Src/test.0008.tif
Each file is a single image.

I run the following command...

Code: Select all

convert C:/Src/test.%04d.tif[4-8] C:/Step1/test.%04d.tif
I now have the following generated files....

Code: Select all

C:/Step1/test.0004.tif
C:/Step1/test.0005.tif
C:/Step1/test.0006.tif
C:/Step1/test.0007.tif
C:/Step1/test.0008.tif
Each file is again a single image.

I run the exact same command on the new files.

Code: Select all

convert C:/Step1/test.%04d.tif[4-8]  C:/Step2/test.%04d.tif
I now have the following...

Code: Select all

C:/Step2/test.0000.tif
C:/Step2/test.0001.tif
C:/Step2/test.0002.tif
C:/Step2/test.0003.tif
C:/Step2/test.0004.tif
I'd like to know why these images only get renamed the second time I run convert.
User avatar
fmw42
Posts: 25562
Joined: 2007-07-02T17:14:51-07:00
Authentication code: 1152
Location: Sunnyvale, California, USA

Re: Frame counter on sequences that don't start at frame 1

Post by fmw42 »

On IM 6.6.3.4 Q16 Mac OSX Tiger:

convert rose: rose_01.tif
convert rose: rose_02.tif
convert rose: rose_03.tif
convert rose: rose_04.tif
convert rose: rose_05.tif
convert rose: rose_06.tif

convert rose_%02d.tif[3-5] rose2_%02d.tif

produces

rose2_00.tif
rose2_01.tif
rose2_02.tif


But


convert rose_%02d.tif[3-5] -scene 3 rose2_%02d.tif

produces

rose2_03.tif
rose2_04.tif
rose2_05.tif


convert rose2_%02d.tif[3-5] rose3_%02d.tif

produces

rose3_00.tif
rose3_01.tif
rose3_02.tif

and

convert rose2_%02d.tif[3-5] -scene 3 rose3_%02d.tif

produces

rose3_03.tif
rose3_04.tif
rose3_05.tif


see http://www.imagemagick.org/Usage/files/#scene


I think your problem is that IM will write starting at 0 if no -scene is specified. When you leave -scene off, it starts with 0 because your second set does not have any images numbered 0000, 0001, 0002, 00003

I think the numbering corresponds to the sequence number of the images in the command line (always starting with 0) and not the numbers in the names of the files.
User avatar
anthony
Posts: 8883
Joined: 2004-05-31T19:27:03-07:00
Authentication code: 8675308
Location: Brisbane, Australia

Re: Frame counter on sequences that don't start at frame 1

Post by anthony »

What is needed is a proper definition of %[scene] %s and %n and any other escapes that may be added,
as well as have it work properly for multi-image reads, and for looped single image operations.
  • For example percent escapes could specify....
  • the scene number READ FROM a multi-image file (related to filename and should be hard to change)
  • the scene for OUTPUT image
  • the image index number in the current image list
All three numbers are related but are also very distinct from each other!

For example you may want to reference the original scene number, but have a different output scene number, which is also different to the images current 'index' for its in memory position.

NOTE you may not have set a 'output scene number' (typically by using -scene) in which case it should default to ... WHAT? typically that should be the current index number at the time of write, as most images would have a input scene number of '0'. It also prevents posible duplicate output scene numbers.

So the bug as I see it is that
  • the first two are current represents %[scene]. They probably should be separated!
  • often %[scene] is getting trashed by the current image index number
  • loop over list image operations are not giving a correct 'index' number or even a correct 'scene' number
    This is especially important for things like -set and -annotate.
  • What the output scene number should be is unclear.
So first lets get the definitions correct, then get the IM to work to those definitions.
Anthony Thyssen -- Webmaster for ImageMagick Example Pages
https://imagemagick.org/Usage/
Post Reply