Netcam Modulo 16 Patch

Uploaded by KennethLavrsen on behalf of PeterHolik who posted it on the mailing list.

Introduction

Motion cannot handle image dimensions unless they are a multiple of 16.

This is normally not a problem because all normal camera image size are multiples of 16.

But when you use a Netcam which may be the jpeg type the image could be scaled to anything.

Today Motion may fail when connected to such an image.

Description of Patch

The patch introduces a check for mulitiple of 16 and

• Logs an error message
• Exits the thread

Installation of Patch

patch < filename

Change History of Patch

Discussion and Comments

The patch came to late to be included in 3.2.2. We want to test it and maybe discuss if this is how we long term want to handle this. Maybe there is a way to make it work?

Patch is put on TWiki so we do not forget it.

-- KennethLavrsen - 13 Aug 2005

---

I took the liberty of uploading the diff file as well.

Two comments about the patch: Thanks - I forgot that - Kenneth

1. I would prefer using % 16 vs. & 0x0F, since the former is much easier to understand in the code. They both result in the same assembler code, so increasing clarity has no performance penalty.
2. Accepting this patch would simplify the rotate feature, since we could use one reverse function always (the fast quad one) instead of choosing one depending on image dimensions.

-- PerJonsson - 13 Aug 2005

---

Patch has been integrated with the following changes.

• 0x0F replaced by 16
• Comments added before the code as suggested in the MotionCodingStandard rule 7
• The random mix of spaces and tabs were change to follow MotionCodingStandard rules 1 and 2.

Will be in 3.2.4_snap1 and in 20050822 daily.

-- KennethLavrsen - 21 Aug 2005

---