I recommend https://welfarefootprint.org/

They do literature reviews, quantify and summarize the main sources of suffering for various farmed animals.

I wrote these two comments about farmed chickens based on their work: https://forum.effectivealtruism.org/posts/3EimeEtikzrTusqYy/animal-welfare-ea-and-personal-dietary-options?commentId=f8MsonDHxHRTvSb25

(Disclaimer: I'm from an animal advocacy group and working in the field for over 10 years.)

Just a point on how the footage from farms is representative, based on your point about not trusting them.

I think you are correct to be skeptical to some of the claims made by documentaries, I feel like some are exaggerating and trying to increase the weight of the claims to make the documentary more appealing. Apart from my personal problem with bending the truth, it's also, I quite confidently think, a bad long-term strategy for the movement. But it highly depends on the filmmaker.

But I really want to note that it's very hard to convey the message to the public about the conditions that animals live in. You may expect that the more brutal footage the better, but it's not the case. We do investigations without knowledge of farm owners (you can check our footage here - https://animainternational.org/resources/investigations - and use it if needed!) and very often we have to use the less inhumane conditions, because our data shows that on average most people are not receptive to faithfully brutal material. It has to be the milder content with enough context for people to sympathize with animals. So you may expect "cherry-picking" in a different direction that you are worried about in terms of them being representative.

There is also an unsurprising problem of not understanding specie welfare needs and animals not showing their suffering in a human's perception compatible way (especially if these are not mammals), so you may see a picture of an animal without any wounds, but it may be in a great suffering because of behavioral needs deprivation (example - repetitive behavior). This is very hard to convey.

So for me, quantitative assessments of suffering between species in farming conditions is the best tool to understand whether animals suffer and to which degree. But I'll add personally that there is an intuition that you get by working with footage/being on farms/working in field that is sometimes hard to capture just by looking at literature (kinda in a similar direction as a point about ground visits when distributing bednets made here). I also wonder how measurement is skewing the results sometimes.

Generally, my bet is that the more data we will get, the more it will show animals suffer more than we expected. My very strong view is that there is sufficient information and it's mostly due to biases that make us discriminate needs of other beings welfare that some people remain undecided on this issue (i.e. we treat interests of non-similar beings to us as less important - animals, future people, digital minds, etc. based on evolutionary heuristics instead of reasoning). That is, unless someone has Yudkowsky's view of sentience with which I strongly disagree (or to be more correct - I disagree with my understanding of that view), but seems logical and coherent to me.

You might be interested in an Adversarial Collaboration I wrote on this topic a few years ago. My collaborator was a meat-eater who was very strong on finding representative statistics (in fact he wrote the first draft of Section 2.2 to keep me extra-honest)

https://slatestarcodex.com/2019/12/11/acc-is-eating-meat-a-net-harm/